rust/src/bootstrap/step.rs

1711 lines
69 KiB
Rust
Raw Normal View History

// Copyright 2016 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
//! Definition of steps of the build system.
//!
//! This is where some of the real meat of rustbuild is located, in how we
//! define targets and the dependencies amongst them. This file can sort of be
//! viewed as just defining targets in a makefile which shell out to predefined
//! functions elsewhere about how to execute the target.
//!
//! The primary function here you're likely interested in is the `build_rules`
//! function. This will create a `Rules` structure which basically just lists
//! everything that rustbuild can do. Each rule has a human-readable name, a
//! path associated with it, some dependencies, and then a closure of how to
//! actually perform the rule.
//!
//! All steps below are defined in self-contained units, so adding a new target
//! to the build system should just involve adding the meta information here
//! along with the actual implementation elsewhere. You can find more comments
//! about how to define rules themselves below.
use std::collections::{BTreeMap, HashSet, HashMap};
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
use std::mem;
use check::{self, TestKind};
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
use compile;
use dist;
use doc;
use flags::Subcommand;
use install;
use native;
use {Compiler, Build, Mode};
pub fn run(build: &Build) {
let rules = build_rules(build);
let steps = rules.plan();
rules.run(&steps);
}
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
pub fn build_rules<'a>(build: &'a Build) -> Rules {
let mut rules = Rules::new(build);
// This is the first rule that we're going to define for rustbuild, which is
// used to compile LLVM itself. All rules are added through the `rules`
// structure created above and are configured through a builder-style
// interface.
//
// First up we see the `build` method. This represents a rule that's part of
// the top-level `build` subcommand. For example `./x.py build` is what this
// is associating with. Note that this is normally only relevant if you flag
// a rule as `default`, which we'll talk about later.
//
// Next up we'll see two arguments to this method:
//
// * `llvm` - this is the "human readable" name of this target. This name is
// not accessed anywhere outside this file itself (e.g. not in
// the CLI nor elsewhere in rustbuild). The purpose of this is to
// easily define dependencies between rules. That is, other rules
// will depend on this with the name "llvm".
// * `src/llvm` - this is the relevant path to the rule that we're working
// with. This path is the engine behind how commands like
// `./x.py build src/llvm` work. This should typically point
// to the relevant component, but if there's not really a
// path to be assigned here you can pass something like
// `path/to/nowhere` to ignore it.
//
// After we create the rule with the `build` method we can then configure
// various aspects of it. For example this LLVM rule uses `.host(true)` to
// flag that it's a rule only for host targets. In other words, LLVM isn't
// compiled for targets configured through `--target` (e.g. those we're just
// building a standard library for).
//
// Next up the `dep` method will add a dependency to this rule. The closure
// is yielded the step that represents executing the `llvm` rule itself
// (containing information like stage, host, target, ...) and then it must
// return a target that the step depends on. Here LLVM is actually
// interesting where a cross-compiled LLVM depends on the host LLVM, but
// otherwise it has no dependencies.
//
// To handle this we do a bit of dynamic dispatch to see what the dependency
// is. If we're building a LLVM for the build triple, then we don't actually
2016-11-16 23:37:20 +01:00
// have any dependencies! To do that we return a dependency on the `Step::noop()`
// target which does nothing.
//
// If we're build a cross-compiled LLVM, however, we need to assemble the
// libraries from the previous compiler. This step has the same name as
// ours (llvm) but we want it for a different target, so we use the
// builder-style methods on `Step` to configure this target to the build
// triple.
//
// Finally, to finish off this rule, we define how to actually execute it.
// That logic is all defined in the `native` module so we just delegate to
// the relevant function there. The argument to the closure passed to `run`
// is a `Step` (defined below) which encapsulates information like the
// stage, target, host, etc.
rules.build("llvm", "src/llvm")
.host(true)
.dep(move |s| {
if s.target == build.config.build {
2016-11-16 23:37:20 +01:00
Step::noop()
} else {
s.target(&build.config.build)
}
})
.run(move |s| native::llvm(build, s.target));
// Ok! After that example rule that's hopefully enough to explain what's
// going on here. You can check out the API docs below and also see a bunch
// more examples of rules directly below as well.
// the compiler with no target libraries ready to go
rules.build("rustc", "src/rustc")
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("create-sysroot").target(s.host))
.dep(move |s| {
if s.stage == 0 {
2016-11-16 23:37:20 +01:00
Step::noop()
} else {
s.name("librustc")
.host(&build.config.build)
.stage(s.stage - 1)
}
})
.run(move |s| compile::assemble_rustc(build, s.stage, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// Helper for loading an entire DAG of crates, rooted at `name`
let krates = |name: &str| {
let mut ret = Vec::new();
let mut list = vec![name];
let mut visited = HashSet::new();
while let Some(krate) = list.pop() {
let default = krate == name;
let krate = &build.crates[krate];
let path = krate.path.strip_prefix(&build.src)
// This handles out of tree paths
.unwrap_or(&krate.path);
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
ret.push((krate, path.to_str().unwrap(), default));
for dep in krate.deps.iter() {
if visited.insert(dep) && dep != "build_helper" {
list.push(dep);
}
}
}
return ret
};
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// ========================================================================
// Crate compilations
//
// Tools used during the build system but not shipped
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
rules.build("create-sysroot", "path/to/nowhere")
.run(move |s| compile::create_sysroot(build, &s.compiler()));
// These rules are "pseudo rules" that don't actually do any work
// themselves, but represent a complete sysroot with the relevant compiler
// linked into place.
//
// That is, depending on "libstd" means that when the rule is completed then
// the `stage` sysroot for the compiler `host` will be available with a
// standard library built for `target` linked in place. Not all rules need
// the compiler itself to be available, just the standard library, so
// there's a distinction between the two.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build("libstd", "src/libstd")
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("rustc").target(s.host))
.dep(|s| s.name("libstd-link"));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build("libtest", "src/libtest")
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libstd"))
.dep(|s| s.name("libtest-link"))
.default(true);
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build("librustc", "src/librustc")
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libtest"))
.dep(|s| s.name("librustc-link"))
.host(true)
.default(true);
// Helper method to define the rules to link a crate into its place in the
// sysroot.
//
// The logic here is a little subtle as there's a few cases to consider.
// Not all combinations of (stage, host, target) actually require something
// to be compiled, but rather libraries could get propagated from a
// different location. For example:
//
// * Any crate with a `host` that's not the build triple will not actually
// compile something. A different `host` means that the build triple will
// actually compile the libraries, and then we'll copy them over from the
// build triple to the `host` directory.
//
// * Some crates aren't even compiled by the build triple, but may be copied
// from previous stages. For example if we're not doing a full bootstrap
// then we may just depend on the stage1 versions of libraries to be
// available to get linked forward.
//
// * Finally, there are some cases, however, which do indeed comiple crates
// and link them into place afterwards.
//
// The rule definition below mirrors these three cases. The `dep` method
// calculates the correct dependency which either comes from stage1, a
// different compiler, or from actually building the crate itself (the `dep`
// rule). The `run` rule then mirrors these three cases and links the cases
// forward into the compiler sysroot specified from the correct location.
fn crate_rule<'a, 'b>(build: &'a Build,
rules: &'b mut Rules<'a>,
krate: &'a str,
dep: &'a str,
link: fn(&Build, &Compiler, &Compiler, &str))
-> RuleBuilder<'a, 'b> {
let mut rule = rules.build(&krate, "path/to/nowhere");
rule.dep(move |s| {
if build.force_use_stage1(&s.compiler(), s.target) {
s.host(&build.config.build).stage(1)
} else if s.host == build.config.build {
s.name(dep)
} else {
s.host(&build.config.build)
}
})
.run(move |s| {
if build.force_use_stage1(&s.compiler(), s.target) {
link(build,
&s.stage(1).host(&build.config.build).compiler(),
&s.compiler(),
s.target)
} else if s.host == build.config.build {
link(build, &s.compiler(), &s.compiler(), s.target)
} else {
link(build,
&s.host(&build.config.build).compiler(),
&s.compiler(),
s.target)
}
});
return rule
}
// Similar to the `libstd`, `libtest`, and `librustc` rules above, except
// these rules only represent the libraries being available in the sysroot,
// not the compiler itself. This is done as not all rules need a compiler in
// the sysroot, but may just need the libraries.
//
// All of these rules use the helper definition above.
crate_rule(build,
&mut rules,
"libstd-link",
"build-crate-std",
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
compile::std_link)
.dep(|s| s.name("startup-objects"))
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("create-sysroot").target(s.host));
crate_rule(build,
&mut rules,
"libtest-link",
"build-crate-test",
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
compile::test_link)
.dep(|s| s.name("libstd-link"));
crate_rule(build,
&mut rules,
"librustc-link",
"build-crate-rustc-main",
compile::rustc_link)
.dep(|s| s.name("libtest-link"));
for (krate, path, _default) in krates("std") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build(&krate.build_step, path)
.dep(|s| s.name("startup-objects"))
2016-11-05 22:22:19 +01:00
.dep(move |s| s.name("rustc").host(&build.config.build).target(s.host))
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.run(move |s| compile::std(build, s.target, &s.compiler()));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
for (krate, path, _default) in krates("test") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build(&krate.build_step, path)
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libstd-link"))
.run(move |s| compile::test(build, s.target, &s.compiler()));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
for (krate, path, _default) in krates("rustc-main") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build(&krate.build_step, path)
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libtest-link"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(move |s| s.name("llvm").host(&build.config.build).stage(0))
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
.dep(|s| s.name("may-run-build-script"))
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.run(move |s| compile::rustc(build, s.target, &s.compiler()));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
// Crates which have build scripts need to rely on this rule to ensure that
// the necessary prerequisites for a build script are linked and located in
// place.
rules.build("may-run-build-script", "path/to/nowhere")
.dep(move |s| {
s.name("libstd-link")
.host(&build.config.build)
.target(&build.config.build)
});
rules.build("startup-objects", "src/rtstartup")
.dep(|s| s.name("create-sysroot").target(s.host))
.run(move |s| compile::build_startup_objects(build, &s.compiler(), s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// ========================================================================
// Test targets
//
// Various unit tests and tests suites we can run
{
let mut suite = |name, path, mode, dir| {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test(name, path)
.dep(|s| s.name("libtest"))
.dep(|s| s.name("tool-compiletest").target(s.host).stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("test-helpers"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
.default(mode != "pretty") // pretty tests don't run everywhere
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| {
check::compiletest(build, &s.compiler(), s.target, mode, dir)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
});
};
suite("check-ui", "src/test/ui", "ui", "ui");
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
suite("check-rpass", "src/test/run-pass", "run-pass", "run-pass");
suite("check-cfail", "src/test/compile-fail", "compile-fail", "compile-fail");
suite("check-pfail", "src/test/parse-fail", "parse-fail", "parse-fail");
suite("check-rfail", "src/test/run-fail", "run-fail", "run-fail");
suite("check-rpass-valgrind", "src/test/run-pass-valgrind",
"run-pass-valgrind", "run-pass-valgrind");
suite("check-mir-opt", "src/test/mir-opt", "mir-opt", "mir-opt");
if build.config.codegen_tests {
suite("check-codegen", "src/test/codegen", "codegen", "codegen");
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
suite("check-codegen-units", "src/test/codegen-units", "codegen-units",
"codegen-units");
suite("check-incremental", "src/test/incremental", "incremental",
"incremental");
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
if build.config.build.contains("msvc") {
// nothing to do for debuginfo tests
} else {
rules.test("check-debuginfo-lldb", "src/test/debuginfo-lldb")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("libtest"))
.dep(|s| s.name("tool-compiletest").target(s.host).stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("test-helpers"))
.dep(|s| s.name("debugger-scripts"))
.run(move |s| check::compiletest(build, &s.compiler(), s.target,
"debuginfo-lldb", "debuginfo"));
rules.test("check-debuginfo-gdb", "src/test/debuginfo-gdb")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("libtest"))
.dep(|s| s.name("tool-compiletest").target(s.host).stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("test-helpers"))
.dep(|s| s.name("debugger-scripts"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::compiletest(build, &s.compiler(), s.target,
"debuginfo-gdb", "debuginfo"));
let mut rule = rules.test("check-debuginfo", "src/test/debuginfo");
rule.default(true);
if build.config.build.contains("apple") {
rule.dep(|s| s.name("check-debuginfo-lldb"));
} else {
rule.dep(|s| s.name("check-debuginfo-gdb"));
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("debugger-scripts", "src/etc/lldb_batchmode.py")
.run(move |s| dist::debugger_scripts(build, &build.sysroot(&s.compiler()),
s.target));
{
let mut suite = |name, path, mode, dir| {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test(name, path)
.dep(|s| s.name("librustc"))
.dep(|s| s.name("test-helpers"))
.dep(|s| s.name("tool-compiletest").target(s.host).stage(0))
.default(mode != "pretty")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.host(true)
.run(move |s| {
check::compiletest(build, &s.compiler(), s.target, mode, dir)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
});
};
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
suite("check-ui-full", "src/test/ui-fulldeps", "ui", "ui-fulldeps");
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
suite("check-rpass-full", "src/test/run-pass-fulldeps",
"run-pass", "run-pass-fulldeps");
suite("check-rfail-full", "src/test/run-fail-fulldeps",
"run-fail", "run-fail-fulldeps");
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
suite("check-cfail-full", "src/test/compile-fail-fulldeps",
"compile-fail", "compile-fail-fulldeps");
suite("check-rmake", "src/test/run-make", "run-make", "run-make");
suite("check-rustdoc", "src/test/rustdoc", "rustdoc", "rustdoc");
suite("check-pretty", "src/test/pretty", "pretty", "pretty");
suite("check-pretty-rpass", "src/test/run-pass/pretty", "pretty",
"run-pass");
suite("check-pretty-rfail", "src/test/run-fail/pretty", "pretty",
"run-fail");
suite("check-pretty-valgrind", "src/test/run-pass-valgrind/pretty", "pretty",
"run-pass-valgrind");
suite("check-pretty-rpass-full", "src/test/run-pass-fulldeps/pretty",
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
"pretty", "run-pass-fulldeps");
suite("check-pretty-rfail-full", "src/test/run-fail-fulldeps/pretty",
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
"pretty", "run-fail-fulldeps");
}
for (krate, path, _default) in krates("std") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test(&krate.test_step, path)
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libstd, TestKind::Test,
Some(&krate.name)));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rules.test("check-std-all", "path/to/nowhere")
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libstd, TestKind::Test, None));
// std benchmarks
for (krate, path, _default) in krates("std") {
rules.bench(&krate.bench_step, path)
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libstd, TestKind::Bench,
Some(&krate.name)));
}
rules.bench("bench-std-all", "path/to/nowhere")
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
.default(true)
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libstd, TestKind::Bench, None));
for (krate, path, _default) in krates("test") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test(&krate.test_step, path)
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libtest, TestKind::Test,
Some(&krate.name)));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rules.test("check-test-all", "path/to/nowhere")
.dep(|s| s.name("libtest"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Libtest, TestKind::Test, None));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
for (krate, path, _default) in krates("rustc-main") {
rules.test(&krate.test_step, path)
2016-11-05 01:44:53 +01:00
.dep(|s| s.name("librustc"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.host(true)
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Librustc, TestKind::Test,
Some(&krate.name)));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
rules.test("check-rustc-all", "path/to/nowhere")
2016-11-05 01:44:53 +01:00
.dep(|s| s.name("librustc"))
2017-04-11 12:10:05 +02:00
.dep(|s| s.name("remote-copy-libs"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
2016-11-05 22:22:19 +01:00
.host(true)
.run(move |s| check::krate(build, &s.compiler(), s.target,
Mode::Librustc, TestKind::Test, None));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("check-linkchecker", "src/tools/linkchecker")
.dep(|s| s.name("tool-linkchecker").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("default:doc"))
.default(true)
2016-11-05 22:22:19 +01:00
.host(true)
.run(move |s| check::linkcheck(build, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("check-cargotest", "src/tools/cargotest")
.dep(|s| s.name("tool-cargotest").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("librustc"))
2016-11-05 22:22:19 +01:00
.host(true)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::cargotest(build, s.stage, s.target));
rules.test("check-cargo", "cargo")
.dep(|s| s.name("tool-cargo"))
.host(true)
.run(move |s| check::cargo(build, s.stage, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("check-tidy", "src/tools/tidy")
.dep(|s| s.name("tool-tidy").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
2016-11-05 22:22:19 +01:00
.host(true)
.only_build(true)
.run(move |s| check::tidy(build, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("check-error-index", "src/tools/error_index_generator")
.dep(|s| s.name("libstd"))
.dep(|s| s.name("tool-error-index").host(s.host).stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
2016-11-05 22:22:19 +01:00
.host(true)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::error_index(build, &s.compiler()));
rules.test("check-docs", "src/doc")
.dep(|s| s.name("libtest"))
.default(true)
2016-11-05 22:22:19 +01:00
.host(true)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| check::docs(build, &s.compiler()));
rules.test("check-distcheck", "distcheck")
.dep(|s| s.name("dist-src"))
.run(move |_| check::distcheck(build));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.build("test-helpers", "src/rt/rust_test_helpers.c")
.run(move |s| native::test_helpers(build, s.target));
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
rules.build("openssl", "path/to/nowhere")
.run(move |s| native::openssl(build, s.target));
2017-04-11 12:10:05 +02:00
// Some test suites are run inside emulators or on remote devices, and most
// of our test binaries are linked dynamically which means we need to ship
// the standard library and such to the emulator ahead of time. This step
// represents this and is a dependency of all test suites.
//
// Most of the time this step is a noop (the `check::emulator_copy_libs`
// only does work if necessary). For some steps such as shipping data to
// QEMU we have to build our own tools so we've got conditional dependencies
2017-04-11 12:10:05 +02:00
// on those programs as well. Note that the remote test client is built for
// the build target (us) and the server is built for the target.
rules.test("remote-copy-libs", "path/to/nowhere")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("libtest"))
.dep(move |s| {
travis: Parallelize tests on Android Currently our slowest test suite on android, run-pass, takes over 5 times longer than the x86_64 component (~400 -> ~2200s). Typically QEMU emulation does indeed add overhead, but not 5x for this kind of workload. One of the slowest parts of the Android process is that *compilation* happens serially. Tests themselves need to run single-threaded on the emulator (due to how the test harness works) and this forces the compiles themselves to be single threaded. Now Travis gives us more than one core per machine, so it'd be much better if we could take advantage of them! The emulator itself is still fundamentally single-threaded, but we should see a nice speedup by sending binaries for it to run much more quickly. It turns out that we've already got all the tools to do this in-tree. The qemu-test-{server,client} that are in use for the ARM Linux testing are a perfect match for the Android emulator. This commit migrates the custom adb management code in compiletest/rustbuild to the same qemu-test-{server,client} implementation that ARM Linux uses. This allows us to lift the parallelism restriction on the compiletest test suites, namely run-pass. Consequently although we'll still basically run the tests themselves in single threaded mode we'll be able to compile all of them in parallel, keeping the pipeline much more full and using more cores for the work at hand. Additionally the architecture here should be a bit speedier as it should have less overhead than adb which is a whole new process on both the host and the emulator! Locally on an 8 core machine I've seen the run-pass test suite speed up from taking nearly an hour to only taking 6 minutes. I don't think we'll see quite a drastic speedup on Travis but I'm hoping this change can place the Android tests well below 2 hours instead of just above 2 hours. Because the client/server here are now repurposed for more than just QEMU, they've been renamed to `remote-test-{server,client}`. Note that this PR does not currently modify how debuginfo tests are executed on Android. While parallelizable it wouldn't be quite as easy, so that's left to another day. Thankfully that test suite is much smaller than the run-pass test suite. As a final fix I discovered that the ARM and Android test suites were actually running all library unit tests (e.g. stdtest, coretest, etc) twice. I've corrected that to only run tests once which should also give a nice boost in overall cycle time here.
2017-04-26 17:52:19 +02:00
if build.remote_tested(s.target) {
s.name("tool-remote-test-client").target(s.host).stage(0)
} else {
Step::noop()
}
})
.dep(move |s| {
travis: Parallelize tests on Android Currently our slowest test suite on android, run-pass, takes over 5 times longer than the x86_64 component (~400 -> ~2200s). Typically QEMU emulation does indeed add overhead, but not 5x for this kind of workload. One of the slowest parts of the Android process is that *compilation* happens serially. Tests themselves need to run single-threaded on the emulator (due to how the test harness works) and this forces the compiles themselves to be single threaded. Now Travis gives us more than one core per machine, so it'd be much better if we could take advantage of them! The emulator itself is still fundamentally single-threaded, but we should see a nice speedup by sending binaries for it to run much more quickly. It turns out that we've already got all the tools to do this in-tree. The qemu-test-{server,client} that are in use for the ARM Linux testing are a perfect match for the Android emulator. This commit migrates the custom adb management code in compiletest/rustbuild to the same qemu-test-{server,client} implementation that ARM Linux uses. This allows us to lift the parallelism restriction on the compiletest test suites, namely run-pass. Consequently although we'll still basically run the tests themselves in single threaded mode we'll be able to compile all of them in parallel, keeping the pipeline much more full and using more cores for the work at hand. Additionally the architecture here should be a bit speedier as it should have less overhead than adb which is a whole new process on both the host and the emulator! Locally on an 8 core machine I've seen the run-pass test suite speed up from taking nearly an hour to only taking 6 minutes. I don't think we'll see quite a drastic speedup on Travis but I'm hoping this change can place the Android tests well below 2 hours instead of just above 2 hours. Because the client/server here are now repurposed for more than just QEMU, they've been renamed to `remote-test-{server,client}`. Note that this PR does not currently modify how debuginfo tests are executed on Android. While parallelizable it wouldn't be quite as easy, so that's left to another day. Thankfully that test suite is much smaller than the run-pass test suite. As a final fix I discovered that the ARM and Android test suites were actually running all library unit tests (e.g. stdtest, coretest, etc) twice. I've corrected that to only run tests once which should also give a nice boost in overall cycle time here.
2017-04-26 17:52:19 +02:00
if build.remote_tested(s.target) {
s.name("tool-remote-test-server")
} else {
Step::noop()
}
})
2017-04-11 12:10:05 +02:00
.run(move |s| check::remote_copy_libs(build, &s.compiler(), s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.test("check-bootstrap", "src/bootstrap")
.default(true)
.host(true)
.only_build(true)
.run(move |_| check::bootstrap(build));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// ========================================================================
// Build tools
//
// Tools used during the build system but not shipped
rules.build("tool-rustbook", "src/tools/rustbook")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("librustc-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "rustbook"));
rules.build("tool-error-index", "src/tools/error_index_generator")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("librustc-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "error_index_generator"));
rules.build("tool-tidy", "src/tools/tidy")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "tidy"));
rules.build("tool-linkchecker", "src/tools/linkchecker")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "linkchecker"));
rules.build("tool-cargotest", "src/tools/cargotest")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "cargotest"));
rules.build("tool-compiletest", "src/tools/compiletest")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libtest-tool"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "compiletest"));
rules.build("tool-build-manifest", "src/tools/build-manifest")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
.run(move |s| compile::tool(build, s.stage, s.target, "build-manifest"));
travis: Parallelize tests on Android Currently our slowest test suite on android, run-pass, takes over 5 times longer than the x86_64 component (~400 -> ~2200s). Typically QEMU emulation does indeed add overhead, but not 5x for this kind of workload. One of the slowest parts of the Android process is that *compilation* happens serially. Tests themselves need to run single-threaded on the emulator (due to how the test harness works) and this forces the compiles themselves to be single threaded. Now Travis gives us more than one core per machine, so it'd be much better if we could take advantage of them! The emulator itself is still fundamentally single-threaded, but we should see a nice speedup by sending binaries for it to run much more quickly. It turns out that we've already got all the tools to do this in-tree. The qemu-test-{server,client} that are in use for the ARM Linux testing are a perfect match for the Android emulator. This commit migrates the custom adb management code in compiletest/rustbuild to the same qemu-test-{server,client} implementation that ARM Linux uses. This allows us to lift the parallelism restriction on the compiletest test suites, namely run-pass. Consequently although we'll still basically run the tests themselves in single threaded mode we'll be able to compile all of them in parallel, keeping the pipeline much more full and using more cores for the work at hand. Additionally the architecture here should be a bit speedier as it should have less overhead than adb which is a whole new process on both the host and the emulator! Locally on an 8 core machine I've seen the run-pass test suite speed up from taking nearly an hour to only taking 6 minutes. I don't think we'll see quite a drastic speedup on Travis but I'm hoping this change can place the Android tests well below 2 hours instead of just above 2 hours. Because the client/server here are now repurposed for more than just QEMU, they've been renamed to `remote-test-{server,client}`. Note that this PR does not currently modify how debuginfo tests are executed on Android. While parallelizable it wouldn't be quite as easy, so that's left to another day. Thankfully that test suite is much smaller than the run-pass test suite. As a final fix I discovered that the ARM and Android test suites were actually running all library unit tests (e.g. stdtest, coretest, etc) twice. I've corrected that to only run tests once which should also give a nice boost in overall cycle time here.
2017-04-26 17:52:19 +02:00
rules.build("tool-remote-test-server", "src/tools/remote-test-server")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
travis: Parallelize tests on Android Currently our slowest test suite on android, run-pass, takes over 5 times longer than the x86_64 component (~400 -> ~2200s). Typically QEMU emulation does indeed add overhead, but not 5x for this kind of workload. One of the slowest parts of the Android process is that *compilation* happens serially. Tests themselves need to run single-threaded on the emulator (due to how the test harness works) and this forces the compiles themselves to be single threaded. Now Travis gives us more than one core per machine, so it'd be much better if we could take advantage of them! The emulator itself is still fundamentally single-threaded, but we should see a nice speedup by sending binaries for it to run much more quickly. It turns out that we've already got all the tools to do this in-tree. The qemu-test-{server,client} that are in use for the ARM Linux testing are a perfect match for the Android emulator. This commit migrates the custom adb management code in compiletest/rustbuild to the same qemu-test-{server,client} implementation that ARM Linux uses. This allows us to lift the parallelism restriction on the compiletest test suites, namely run-pass. Consequently although we'll still basically run the tests themselves in single threaded mode we'll be able to compile all of them in parallel, keeping the pipeline much more full and using more cores for the work at hand. Additionally the architecture here should be a bit speedier as it should have less overhead than adb which is a whole new process on both the host and the emulator! Locally on an 8 core machine I've seen the run-pass test suite speed up from taking nearly an hour to only taking 6 minutes. I don't think we'll see quite a drastic speedup on Travis but I'm hoping this change can place the Android tests well below 2 hours instead of just above 2 hours. Because the client/server here are now repurposed for more than just QEMU, they've been renamed to `remote-test-{server,client}`. Note that this PR does not currently modify how debuginfo tests are executed on Android. While parallelizable it wouldn't be quite as easy, so that's left to another day. Thankfully that test suite is much smaller than the run-pass test suite. As a final fix I discovered that the ARM and Android test suites were actually running all library unit tests (e.g. stdtest, coretest, etc) twice. I've corrected that to only run tests once which should also give a nice boost in overall cycle time here.
2017-04-26 17:52:19 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "remote-test-server"));
rules.build("tool-remote-test-client", "src/tools/remote-test-client")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
travis: Parallelize tests on Android Currently our slowest test suite on android, run-pass, takes over 5 times longer than the x86_64 component (~400 -> ~2200s). Typically QEMU emulation does indeed add overhead, but not 5x for this kind of workload. One of the slowest parts of the Android process is that *compilation* happens serially. Tests themselves need to run single-threaded on the emulator (due to how the test harness works) and this forces the compiles themselves to be single threaded. Now Travis gives us more than one core per machine, so it'd be much better if we could take advantage of them! The emulator itself is still fundamentally single-threaded, but we should see a nice speedup by sending binaries for it to run much more quickly. It turns out that we've already got all the tools to do this in-tree. The qemu-test-{server,client} that are in use for the ARM Linux testing are a perfect match for the Android emulator. This commit migrates the custom adb management code in compiletest/rustbuild to the same qemu-test-{server,client} implementation that ARM Linux uses. This allows us to lift the parallelism restriction on the compiletest test suites, namely run-pass. Consequently although we'll still basically run the tests themselves in single threaded mode we'll be able to compile all of them in parallel, keeping the pipeline much more full and using more cores for the work at hand. Additionally the architecture here should be a bit speedier as it should have less overhead than adb which is a whole new process on both the host and the emulator! Locally on an 8 core machine I've seen the run-pass test suite speed up from taking nearly an hour to only taking 6 minutes. I don't think we'll see quite a drastic speedup on Travis but I'm hoping this change can place the Android tests well below 2 hours instead of just above 2 hours. Because the client/server here are now repurposed for more than just QEMU, they've been renamed to `remote-test-{server,client}`. Note that this PR does not currently modify how debuginfo tests are executed on Android. While parallelizable it wouldn't be quite as easy, so that's left to another day. Thankfully that test suite is much smaller than the run-pass test suite. As a final fix I discovered that the ARM and Android test suites were actually running all library unit tests (e.g. stdtest, coretest, etc) twice. I've corrected that to only run tests once which should also give a nice boost in overall cycle time here.
2017-04-26 17:52:19 +02:00
.run(move |s| compile::tool(build, s.stage, s.target, "remote-test-client"));
2017-05-09 00:01:13 +02:00
rules.build("tool-rust-installer", "src/tools/rust-installer")
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
.run(move |s| compile::tool(build, s.stage, s.target, "rust-installer"));
rules.build("tool-cargo", "src/tools/cargo")
.host(true)
.default(build.config.extended)
.dep(|s| s.name("maybe-clean-tools"))
.dep(|s| s.name("libstd-tool"))
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
.dep(|s| s.stage(0).host(s.target).name("openssl"))
.dep(move |s| {
// Cargo depends on procedural macros, which requires a full host
// compiler to be available, so we need to depend on that.
s.name("librustc-link")
.target(&build.config.build)
.host(&build.config.build)
})
.run(move |s| compile::tool(build, s.stage, s.target, "cargo"));
rules.build("tool-rls", "src/tools/rls")
.host(true)
.default(build.config.extended)
.dep(|s| s.name("librustc-tool"))
.dep(|s| s.stage(0).host(s.target).name("openssl"))
.dep(move |s| {
// rls, like cargo, uses procedural macros
s.name("librustc-link")
.target(&build.config.build)
.host(&build.config.build)
})
.run(move |s| compile::tool(build, s.stage, s.target, "rls"));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// "pseudo rule" which represents completely cleaning out the tools dir in
// one stage. This needs to happen whenever a dependency changes (e.g.
// libstd, libtest, librustc) and all of the tool compilations above will
// be sequenced after this rule.
rules.build("maybe-clean-tools", "path/to/nowhere")
.after("librustc-tool")
.after("libtest-tool")
.after("libstd-tool");
rules.build("librustc-tool", "path/to/nowhere")
.dep(|s| s.name("librustc"))
.run(move |s| compile::maybe_clean_tools(build, s.stage, s.target, Mode::Librustc));
rules.build("libtest-tool", "path/to/nowhere")
.dep(|s| s.name("libtest"))
.run(move |s| compile::maybe_clean_tools(build, s.stage, s.target, Mode::Libtest));
rules.build("libstd-tool", "path/to/nowhere")
.dep(|s| s.name("libstd"))
.run(move |s| compile::maybe_clean_tools(build, s.stage, s.target, Mode::Libstd));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// ========================================================================
// Documentation targets
rules.doc("doc-book", "src/doc/book")
.dep(move |s| {
s.name("tool-rustbook")
.host(&build.config.build)
.target(&build.config.build)
.stage(0)
})
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(build.config.docs)
2017-03-07 20:49:50 +01:00
.run(move |s| doc::book(build, s.target, "book"));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.doc("doc-nomicon", "src/doc/nomicon")
.dep(move |s| {
s.name("tool-rustbook")
.host(&build.config.build)
.target(&build.config.build)
.stage(0)
})
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(build.config.docs)
.run(move |s| doc::rustbook(build, s.target, "nomicon"));
rules.doc("doc-reference", "src/doc/reference")
.dep(move |s| {
s.name("tool-rustbook")
.host(&build.config.build)
.target(&build.config.build)
.stage(0)
})
.default(build.config.docs)
.run(move |s| doc::rustbook(build, s.target, "reference"));
rules.doc("doc-unstable-book", "src/doc/unstable-book")
.dep(move |s| {
s.name("tool-rustbook")
.host(&build.config.build)
.target(&build.config.build)
.stage(0)
})
.default(build.config.docs)
.run(move |s| doc::rustbook(build, s.target, "unstable-book"));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.doc("doc-standalone", "src/doc")
.dep(move |s| {
s.name("rustc")
.host(&build.config.build)
.target(&build.config.build)
.stage(0)
})
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(build.config.docs)
.run(move |s| doc::standalone(build, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.doc("doc-error-index", "src/tools/error_index_generator")
.dep(move |s| s.name("tool-error-index").target(&build.config.build).stage(0))
.dep(move |s| s.name("librustc-link"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(build.config.docs)
2016-11-05 22:22:19 +01:00
.host(true)
.run(move |s| doc::error_index(build, s.target));
for (krate, path, default) in krates("std") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.doc(&krate.doc_step, path)
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libstd-link"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(default && build.config.docs)
.run(move |s| doc::std(build, s.stage, s.target));
}
for (krate, path, default) in krates("test") {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.doc(&krate.doc_step, path)
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("libtest-link"))
// Needed so rustdoc generates relative links to std.
.dep(|s| s.name("doc-crate-std"))
.default(default && build.config.compiler_docs)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| doc::test(build, s.stage, s.target));
}
for (krate, path, default) in krates("rustc-main") {
rules.doc(&krate.doc_step, path)
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
.dep(|s| s.name("librustc-link"))
// Needed so rustdoc generates relative links to std.
.dep(|s| s.name("doc-crate-std"))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.host(true)
.default(default && build.config.docs)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| doc::rustc(build, s.stage, s.target));
}
// ========================================================================
// Distribution targets
rules.dist("dist-rustc", "src/librustc")
2016-11-05 22:22:19 +01:00
.dep(move |s| s.name("rustc").host(&build.config.build))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.host(true)
.only_host_build(true)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.default(true)
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| dist::rustc(build, s.stage, s.target));
rules.dist("dist-std", "src/libstd")
.dep(move |s| {
// We want to package up as many target libraries as possible
// for the `rust-std` package, so if this is a host target we
// depend on librustc and otherwise we just depend on libtest.
if build.config.host.iter().any(|t| t == s.target) {
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
s.name("librustc-link")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
} else {
rustbuild: Compile rustc twice, not thrice This commit switches the rustbuild build system to compiling the compiler twice for a normal bootstrap rather than the historical three times. Rust is a bootstrapped language which means that a previous version of the compiler is used to build the next version of the compiler. Over time, however, we change many parts of compiler artifacts such as the metadata format, symbol names, etc. These changes make artifacts from one compiler incompatible from another compiler. Consequently if a compiler wants to be able to use some artifacts then it itself must have compiled the artifacts. Historically the rustc build system has achieved this by compiling the compiler three times: * An older compiler (stage0) is downloaded to kick off the chain. * This compiler now compiles a new compiler (stage1) * The stage1 compiler then compiles another compiler (stage2) * Finally, the stage2 compiler needs libraries to link against, so it compiles all the libraries again. This entire process amounts in compiling the compiler three times. Additionally, this process always guarantees that the Rust source tree can compile itself because the stage2 compiler (created by a freshly created compiler) would successfully compile itself again. This property, ensuring Rust can compile itself, is quite important! In general, though, this third compilation is not required for general purpose development on the compiler. The third compiler (stage2) can reuse the libraries that were created during the second compile. In other words, the second compilation can produce both a compiler and the libraries that compiler will use. These artifacts *must* be compatible due to the way plugins work today anyway, and they were created by the same source code so they *should* be compatible as well. So given all that, this commit switches the default build process to only compile the compiler three times, avoiding this third compilation by copying artifacts from the previous one. Along the way a new entry in the Travis matrix was also added to ensure that our full bootstrap can succeed. This entry does not run tests, though, as it should not be necessary. To restore the old behavior of a full bootstrap (three compiles) you can either pass: ./configure --enable-full-bootstrap or if you're using config.toml: [build] full-bootstrap = true Overall this will hopefully be an easy 33% win in build times of the compiler. If we do 33% less work we should be 33% faster! This in turn should affect cycle times and such on Travis and AppVeyor positively as well as making it easier to work on the compiler itself.
2016-12-26 00:20:33 +01:00
s.name("libtest-link")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
})
.default(true)
.only_host_build(true)
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| dist::std(build, &s.compiler(), s.target));
rules.dist("dist-mingw", "path/to/nowhere")
.default(true)
.only_host_build(true)
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |s| {
if s.target.contains("pc-windows-gnu") {
dist::mingw(build, s.target)
}
});
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.dist("dist-src", "src")
.default(true)
.host(true)
.only_build(true)
.only_host_build(true)
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |_| dist::rust_src(build));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.dist("dist-docs", "src/doc")
.default(true)
.only_host_build(true)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("default:doc"))
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.run(move |s| dist::docs(build, s.stage, s.target));
rules.dist("dist-analysis", "analysis")
.default(build.config.extended)
.dep(|s| s.name("dist-std"))
.only_host_build(true)
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |s| dist::analysis(build, &s.compiler(), s.target));
rules.dist("dist-rls", "rls")
.host(true)
.only_host_build(true)
.dep(|s| s.name("tool-rls"))
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |s| dist::rls(build, s.stage, s.target));
rules.dist("install", "path/to/nowhere")
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
.dep(|s| s.name("default:dist"))
.run(move |s| install::install(build, s.stage, s.target));
rules.dist("dist-cargo", "cargo")
.host(true)
.only_host_build(true)
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
.dep(|s| s.name("tool-cargo"))
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |s| dist::cargo(build, s.stage, s.target));
rules.dist("dist-extended", "extended")
.default(build.config.extended)
.host(true)
.only_host_build(true)
.dep(|d| d.name("dist-std"))
.dep(|d| d.name("dist-rustc"))
.dep(|d| d.name("dist-mingw"))
.dep(|d| d.name("dist-docs"))
.dep(|d| d.name("dist-cargo"))
.dep(|d| d.name("dist-rls"))
.dep(|d| d.name("dist-analysis"))
2017-05-09 00:01:13 +02:00
.dep(|s| s.name("tool-rust-installer").stage(0))
.run(move |s| dist::extended(build, s.stage, s.target));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.dist("dist-sign", "hash-and-sign")
.host(true)
.only_build(true)
.only_host_build(true)
.dep(move |s| s.name("tool-build-manifest").target(&build.config.build).stage(0))
.run(move |_| dist::hash_and_sign(build));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
rules.verify();
return rules;
}
#[derive(PartialEq, Eq, Hash, Clone, Debug)]
struct Step<'a> {
/// Human readable name of the rule this step is executing. Possible names
/// are all defined above in `build_rules`.
name: &'a str,
/// The stage this step is executing in. This is typically 0, 1, or 2.
stage: u32,
/// This step will likely involve a compiler, and the target that compiler
/// itself is built for is called the host, this variable. Typically this is
/// the target of the build machine itself.
host: &'a str,
/// The target that this step represents generating. If you're building a
/// standard library for a new suite of targets, for example, this'll be set
/// to those targets.
target: &'a str,
}
impl<'a> Step<'a> {
2016-11-16 23:37:20 +01:00
fn noop() -> Step<'a> {
Step { name: "", stage: 0, host: "", target: "" }
}
/// Creates a new step which is the same as this, except has a new name.
fn name(&self, name: &'a str) -> Step<'a> {
Step { name: name, ..*self }
}
/// Creates a new step which is the same as this, except has a new stage.
fn stage(&self, stage: u32) -> Step<'a> {
Step { stage: stage, ..*self }
}
/// Creates a new step which is the same as this, except has a new host.
fn host(&self, host: &'a str) -> Step<'a> {
Step { host: host, ..*self }
}
/// Creates a new step which is the same as this, except has a new target.
fn target(&self, target: &'a str) -> Step<'a> {
Step { target: target, ..*self }
}
/// Returns the `Compiler` structure that this step corresponds to.
fn compiler(&self) -> Compiler<'a> {
Compiler::new(self.stage, self.host)
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
struct Rule<'a> {
/// The human readable name of this target, defined in `build_rules`.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
name: &'a str,
/// The path associated with this target, used in the `./x.py` driver for
/// easy and ergonomic specification of what to do.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
path: &'a str,
/// The "kind" of top-level command that this rule is associated with, only
/// relevant if this is a default rule.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
kind: Kind,
/// List of dependencies this rule has. Each dependency is a function from a
/// step that's being executed to another step that should be executed.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
deps: Vec<Box<Fn(&Step<'a>) -> Step<'a> + 'a>>,
/// How to actually execute this rule. Takes a step with contextual
/// information and then executes it.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
run: Box<Fn(&Step<'a>) + 'a>,
/// Whether or not this is a "default" rule. That basically means that if
/// you run, for example, `./x.py test` whether it's included or not.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
default: bool,
/// Whether or not this is a "host" rule, or in other words whether this is
/// only intended for compiler hosts and not for targets that are being
/// generated.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
host: bool,
/// Whether this rule is only for steps where the host is the build triple,
/// not anything in hosts or targets.
only_host_build: bool,
/// Whether this rule is only for the build triple, not anything in hosts or
/// targets.
only_build: bool,
/// A list of "order only" dependencies. This rules does not actually
/// depend on these rules, but if they show up in the dependency graph then
/// this rule must be executed after all these rules.
after: Vec<&'a str>,
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
#[derive(PartialEq)]
enum Kind {
Build,
Test,
Bench,
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
Dist,
Doc,
}
impl<'a> Rule<'a> {
fn new(name: &'a str, path: &'a str, kind: Kind) -> Rule<'a> {
Rule {
name: name,
deps: Vec::new(),
run: Box::new(|_| ()),
path: path,
kind: kind,
default: false,
host: false,
only_host_build: false,
only_build: false,
after: Vec::new(),
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
}
/// Builder pattern returned from the various methods on `Rules` which will add
/// the rule to the internal list on `Drop`.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
struct RuleBuilder<'a: 'b, 'b> {
rules: &'b mut Rules<'a>,
rule: Rule<'a>,
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
impl<'a, 'b> RuleBuilder<'a, 'b> {
fn dep<F>(&mut self, f: F) -> &mut Self
where F: Fn(&Step<'a>) -> Step<'a> + 'a,
{
self.rule.deps.push(Box::new(f));
self
}
fn after(&mut self, step: &'a str) -> &mut Self {
self.rule.after.push(step);
self
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn run<F>(&mut self, f: F) -> &mut Self
where F: Fn(&Step<'a>) + 'a,
{
self.rule.run = Box::new(f);
self
}
fn default(&mut self, default: bool) -> &mut Self {
self.rule.default = default;
self
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn host(&mut self, host: bool) -> &mut Self {
self.rule.host = host;
self
}
fn only_build(&mut self, only_build: bool) -> &mut Self {
self.rule.only_build = only_build;
self
}
fn only_host_build(&mut self, only_host_build: bool) -> &mut Self {
self.rule.only_host_build = only_host_build;
self
}
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
impl<'a, 'b> Drop for RuleBuilder<'a, 'b> {
fn drop(&mut self) {
let rule = mem::replace(&mut self.rule, Rule::new("", "", Kind::Build));
let prev = self.rules.rules.insert(rule.name, rule);
if let Some(prev) = prev {
panic!("duplicate rule named: {}", prev.name);
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
pub struct Rules<'a> {
build: &'a Build,
sbuild: Step<'a>,
rules: BTreeMap<&'a str, Rule<'a>>,
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
impl<'a> Rules<'a> {
fn new(build: &'a Build) -> Rules<'a> {
Rules {
build: build,
sbuild: Step {
stage: build.flags.stage.unwrap_or(2),
target: &build.config.build,
host: &build.config.build,
name: "",
},
rules: BTreeMap::new(),
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
}
/// Creates a new rule of `Kind::Build` with the specified human readable
/// name and path associated with it.
///
/// The builder returned should be configured further with information such
/// as how to actually run this rule.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn build<'b>(&'b mut self, name: &'a str, path: &'a str)
-> RuleBuilder<'a, 'b> {
self.rule(name, path, Kind::Build)
}
/// Same as `build`, but for `Kind::Test`.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn test<'b>(&'b mut self, name: &'a str, path: &'a str)
-> RuleBuilder<'a, 'b> {
self.rule(name, path, Kind::Test)
}
/// Same as `build`, but for `Kind::Bench`.
fn bench<'b>(&'b mut self, name: &'a str, path: &'a str)
-> RuleBuilder<'a, 'b> {
self.rule(name, path, Kind::Bench)
}
/// Same as `build`, but for `Kind::Doc`.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn doc<'b>(&'b mut self, name: &'a str, path: &'a str)
-> RuleBuilder<'a, 'b> {
self.rule(name, path, Kind::Doc)
}
/// Same as `build`, but for `Kind::Dist`.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn dist<'b>(&'b mut self, name: &'a str, path: &'a str)
-> RuleBuilder<'a, 'b> {
self.rule(name, path, Kind::Dist)
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
fn rule<'b>(&'b mut self,
name: &'a str,
path: &'a str,
kind: Kind) -> RuleBuilder<'a, 'b> {
RuleBuilder {
rules: self,
rule: Rule::new(name, path, kind),
}
}
/// Verify the dependency graph defined by all our rules are correct, e.g.
/// everything points to a valid something else.
fn verify(&self) {
for rule in self.rules.values() {
for dep in rule.deps.iter() {
let dep = dep(&self.sbuild.name(rule.name));
if self.rules.contains_key(&dep.name) || dep.name.starts_with("default:") {
continue
}
2016-11-16 23:37:20 +01:00
if dep == Step::noop() {
continue
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
panic!("\
invalid rule dependency graph detected, was a rule added and maybe typo'd?
`{}` depends on `{}` which does not exist
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
", rule.name, dep.name);
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
}
pub fn get_help(&self, command: &str) -> Option<String> {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
let kind = match command {
"build" => Kind::Build,
"doc" => Kind::Doc,
"test" => Kind::Test,
"bench" => Kind::Bench,
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
"dist" => Kind::Dist,
_ => return None,
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
};
let rules = self.rules.values().filter(|r| r.kind == kind);
let rules = rules.filter(|r| !r.path.contains("nowhere"));
let mut rules = rules.collect::<Vec<_>>();
rules.sort_by_key(|r| r.path);
let mut help_string = String::from("Available paths:\n");
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
for rule in rules {
help_string.push_str(format!(" ./x.py {} {}\n", command, rule.path).as_str());
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
Some(help_string)
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
/// Construct the top-level build steps that we're going to be executing,
/// given the subcommand that our build is performing.
fn plan(&self) -> Vec<Step<'a>> {
// Ok, the logic here is pretty subtle, and involves quite a few
// conditionals. The basic idea here is to:
//
// 1. First, filter all our rules to the relevant ones. This means that
// the command specified corresponds to one of our `Kind` variants,
// and we filter all rules based on that.
//
// 2. Next, we determine which rules we're actually executing. If a
// number of path filters were specified on the command line we look
// for those, otherwise we look for anything tagged `default`.
// Here we also compute the priority of each rule based on how early
// in the command line the matching path filter showed up.
//
// 3. Finally, we generate some steps with host and target information.
//
// The last step is by far the most complicated and subtle. The basic
// thinking here is that we want to take the cartesian product of
// specified hosts and targets and build rules with that. The list of
// hosts and targets, if not specified, come from the how this build was
// configured. If the rule we're looking at is a host-only rule the we
// ignore the list of targets and instead consider the list of hosts
// also the list of targets.
//
// Once the host and target lists are generated we take the cartesian
// product of the two and then create a step based off them. Note that
// the stage each step is associated was specified with the `--step`
// flag on the command line.
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
let (kind, paths) = match self.build.flags.cmd {
Subcommand::Build { ref paths } => (Kind::Build, &paths[..]),
Subcommand::Doc { ref paths } => (Kind::Doc, &paths[..]),
Subcommand::Test { ref paths, test_args: _ } => (Kind::Test, &paths[..]),
Subcommand::Bench { ref paths, test_args: _ } => (Kind::Bench, &paths[..]),
Subcommand::Dist { ref paths, install } => {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
if install {
return vec![self.sbuild.name("install")]
} else {
(Kind::Dist, &paths[..])
}
}
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
Subcommand::Clean => panic!(),
};
let mut rules: Vec<_> = self.rules.values().filter_map(|rule| {
if rule.kind != kind {
return None;
}
if paths.len() == 0 && rule.default {
Some((rule, 0))
} else {
paths.iter().position(|path| path.ends_with(rule.path))
.map(|priority| (rule, priority))
}
}).collect();
rules.sort_by_key(|&(_, priority)| priority);
rules.into_iter().flat_map(|(rule, _)| {
let hosts = if rule.only_host_build || rule.only_build {
&self.build.config.host[..1]
} else if self.build.flags.host.len() > 0 {
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
&self.build.flags.host
} else {
&self.build.config.host
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
};
let targets = if self.build.flags.target.len() > 0 {
&self.build.flags.target
} else {
&self.build.config.target
};
// Determine the actual targets participating in this rule.
// NOTE: We should keep the full projection from build triple to
// the hosts for the dist steps, now that the hosts array above is
// truncated to avoid duplication of work in that case. Therefore
// the original non-shadowed hosts array is used below.
let arr = if rule.host {
// If --target was specified but --host wasn't specified,
// don't run any host-only tests. Also, respect any `--host`
// overrides as done for `hosts`.
if self.build.flags.host.len() > 0 {
&self.build.flags.host[..]
} else if self.build.flags.target.len() > 0 {
&[]
} else if rule.only_build {
&self.build.config.host[..1]
} else {
&self.build.config.host[..]
}
} else {
targets
};
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
hosts.iter().flat_map(move |host| {
arr.iter().map(move |target| {
self.sbuild.name(rule.name).target(target).host(host)
})
})
}).collect()
}
/// Execute all top-level targets indicated by `steps`.
///
/// This will take the list returned by `plan` and then execute each step
/// along with all required dependencies as it goes up the chain.
fn run(&self, steps: &[Step<'a>]) {
self.build.verbose("bootstrap top targets:");
for step in steps.iter() {
self.build.verbose(&format!("\t{:?}", step));
}
// Using `steps` as the top-level targets, make a topological ordering
// of what we need to do.
let order = self.expand(steps);
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
// Print out what we're doing for debugging
self.build.verbose("bootstrap build plan:");
for step in order.iter() {
self.build.verbose(&format!("\t{:?}", step));
}
// And finally, iterate over everything and execute it.
for step in order.iter() {
if self.build.flags.keep_stage.map_or(false, |s| step.stage <= s) {
self.build.verbose(&format!("keeping step {:?}", step));
continue;
}
2016-11-05 22:22:19 +01:00
self.build.verbose(&format!("executing step {:?}", step));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
(self.rules[step.name].run)(step);
}
}
/// From the top level targets `steps` generate a topological ordering of
/// all steps needed to run those steps.
fn expand(&self, steps: &[Step<'a>]) -> Vec<Step<'a>> {
// First up build a graph of steps and their dependencies. The `nodes`
// map is a map from step to a unique number. The `edges` map is a
// map from these unique numbers to a list of other numbers,
// representing dependencies.
let mut nodes = HashMap::new();
nodes.insert(Step::noop(), 0);
let mut edges = HashMap::new();
edges.insert(0, HashSet::new());
for step in steps {
self.build_graph(step.clone(), &mut nodes, &mut edges);
}
// Now that we've built up the actual dependency graph, draw more
// dependency edges to satisfy the `after` dependencies field for each
// rule.
self.satisfy_after_deps(&nodes, &mut edges);
// And finally, perform a topological sort to return a list of steps to
// execute.
let mut order = Vec::new();
let mut visited = HashSet::new();
visited.insert(0);
let idx_to_node = nodes.iter().map(|p| (*p.1, p.0)).collect::<HashMap<_, _>>();
for idx in 0..nodes.len() {
self.topo_sort(idx, &idx_to_node, &edges, &mut visited, &mut order);
}
return order
}
/// Builds the dependency graph rooted at `step`.
///
/// The `nodes` and `edges` maps are filled out according to the rule
/// described by `step.name`.
fn build_graph(&self,
step: Step<'a>,
nodes: &mut HashMap<Step<'a>, usize>,
edges: &mut HashMap<usize, HashSet<usize>>) -> usize {
use std::collections::hash_map::Entry;
let idx = nodes.len();
match nodes.entry(step.clone()) {
Entry::Vacant(e) => { e.insert(idx); }
Entry::Occupied(e) => return *e.get(),
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
let mut deps = Vec::new();
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
for dep in self.rules[step.name].deps.iter() {
let dep = dep(&step);
if dep.name.starts_with("default:") {
let kind = match &dep.name[8..] {
"doc" => Kind::Doc,
"dist" => Kind::Dist,
kind => panic!("unknown kind: `{}`", kind),
};
let host = self.build.config.host.iter().any(|h| h == dep.target);
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
let rules = self.rules.values().filter(|r| r.default);
for rule in rules.filter(|r| r.kind == kind && (!r.host || host)) {
deps.push(self.build_graph(dep.name(rule.name), nodes, edges));
rustbuild: Rewrite user-facing interface This commit is a rewrite of the user-facing interface to the rustbuild build system. The intention here is to make it much easier to compile/test the project without having to remember weird rule names and such. An overall view of the new interface is: # build everything ./x.py build # document everyting ./x.py doc # test everything ./x.py test # test libstd ./x.py test src/libstd # build libcore stage0 ./x.py build src/libcore --stage 0 # run stage1 run-pass tests ./x.py test src/test/run-pass --stage 1 The `src/bootstrap/bootstrap.py` script is now aliased as a top-level `x.py` script. This `x` was chosen to be both short and easily tab-completable (no collisions in that namespace!). The build system now accepts a "subcommand" of what to do next, the main ones being build/doc/test. Each subcommand then receives an optional list of arguments. These arguments are paths in the source repo of what to work with. That is, if you want to test a directory, you just pass that directory as an argument. The purpose of this rewrite is to do away with all of the arcane renames like "rpass" is the "run-pass" suite, "cfail" is the "compile-fail" suite, etc. By simply working with directories and files it's much more intuitive of how to run a test (just pass it as an argument). The rustbuild step/dependency management was also rewritten along the way to make this easy to work with and define, but that's largely just a refactoring of what was there before. The *intention* is that this support is extended for arbitrary files (e.g. `src/test/run-pass/my-test-case.rs`), but that isn't quite implemented just yet. Instead directories work for now but we can follow up with stricter path filtering logic to plumb through all the arguments.
2016-10-21 22:18:09 +02:00
}
} else {
deps.push(self.build_graph(dep, nodes, edges));
}
}
edges.entry(idx).or_insert(HashSet::new()).extend(deps);
return idx
}
/// Given a dependency graph with a finished list of `nodes`, fill out more
/// dependency `edges`.
///
/// This is the step which satisfies all `after` listed dependencies in
/// `Rule` above.
fn satisfy_after_deps(&self,
nodes: &HashMap<Step<'a>, usize>,
edges: &mut HashMap<usize, HashSet<usize>>) {
// Reverse map from the name of a step to the node indices that it
// appears at.
let mut name_to_idx = HashMap::new();
for (step, &idx) in nodes {
name_to_idx.entry(step.name).or_insert(Vec::new()).push(idx);
}
for (step, idx) in nodes {
if *step == Step::noop() {
continue
}
for after in self.rules[step.name].after.iter() {
// This is the critical piece of an `after` dependency. If the
// dependency isn't actually in our graph then no edge is drawn,
// only if it's already present do we draw the edges.
if let Some(idxs) = name_to_idx.get(after) {
edges.get_mut(idx).unwrap()
.extend(idxs.iter().cloned());
}
}
}
}
fn topo_sort(&self,
cur: usize,
nodes: &HashMap<usize, &Step<'a>>,
edges: &HashMap<usize, HashSet<usize>>,
visited: &mut HashSet<usize>,
order: &mut Vec<Step<'a>>) {
if !visited.insert(cur) {
return
}
for dep in edges[&cur].iter() {
self.topo_sort(*dep, nodes, edges, visited, order);
}
order.push(nodes[&cur].clone());
}
}
#[cfg(test)]
mod tests {
use std::env;
use Build;
use config::Config;
use flags::Flags;
macro_rules! a {
($($a:expr),*) => (vec![$($a.to_string()),*])
}
fn build(args: &[&str],
extra_host: &[&str],
extra_target: &[&str]) -> Build {
let mut args = args.iter().map(|s| s.to_string()).collect::<Vec<_>>();
args.push("--build".to_string());
args.push("A".to_string());
let flags = Flags::parse(&args);
let mut config = Config::default();
config.docs = true;
config.build = "A".to_string();
config.host = vec![config.build.clone()];
config.host.extend(extra_host.iter().map(|s| s.to_string()));
config.target = config.host.clone();
config.target.extend(extra_target.iter().map(|s| s.to_string()));
let mut build = Build::new(flags, config);
let cwd = env::current_dir().unwrap();
build.crates.insert("std".to_string(), ::Crate {
name: "std".to_string(),
deps: Vec::new(),
path: cwd.join("src/std"),
doc_step: "doc-crate-std".to_string(),
build_step: "build-crate-std".to_string(),
test_step: "test-crate-std".to_string(),
bench_step: "bench-crate-std".to_string(),
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
version: String::new(),
});
build.crates.insert("test".to_string(), ::Crate {
name: "test".to_string(),
deps: Vec::new(),
path: cwd.join("src/test"),
doc_step: "doc-crate-test".to_string(),
build_step: "build-crate-test".to_string(),
test_step: "test-crate-test".to_string(),
bench_step: "bench-crate-test".to_string(),
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
version: String::new(),
});
build.crates.insert("rustc-main".to_string(), ::Crate {
name: "rustc-main".to_string(),
deps: Vec::new(),
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
version: String::new(),
path: cwd.join("src/rustc-main"),
doc_step: "doc-crate-rustc-main".to_string(),
build_step: "build-crate-rustc-main".to_string(),
test_step: "test-crate-rustc-main".to_string(),
bench_step: "bench-crate-rustc-main".to_string(),
});
return build
}
#[test]
fn dist_baseline() {
let build = build(&["dist"], &[], &[]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
assert!(plan.iter().all(|s| s.host == "A" ));
assert!(plan.iter().all(|s| s.target == "A" ));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(plan.contains(&step.name("dist-docs")));
assert!(plan.contains(&step.name("dist-mingw")));
assert!(plan.contains(&step.name("dist-rustc")));
assert!(plan.contains(&step.name("dist-std")));
assert!(plan.contains(&step.name("dist-src")));
}
#[test]
fn dist_with_targets() {
let build = build(&["dist"], &[], &["B"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
assert!(plan.iter().all(|s| s.host == "A" ));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(plan.contains(&step.name("dist-docs")));
assert!(plan.contains(&step.name("dist-mingw")));
assert!(plan.contains(&step.name("dist-rustc")));
assert!(plan.contains(&step.name("dist-std")));
assert!(plan.contains(&step.name("dist-src")));
assert!(plan.contains(&step.target("B").name("dist-docs")));
assert!(plan.contains(&step.target("B").name("dist-mingw")));
assert!(!plan.contains(&step.target("B").name("dist-rustc")));
assert!(plan.contains(&step.target("B").name("dist-std")));
assert!(!plan.contains(&step.target("B").name("dist-src")));
}
#[test]
fn dist_with_hosts() {
let build = build(&["dist"], &["B"], &[]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(!plan.iter().any(|s| s.host == "B"));
assert!(plan.contains(&step.name("dist-docs")));
assert!(plan.contains(&step.name("dist-mingw")));
assert!(plan.contains(&step.name("dist-rustc")));
assert!(plan.contains(&step.name("dist-std")));
assert!(plan.contains(&step.name("dist-src")));
assert!(plan.contains(&step.target("B").name("dist-docs")));
assert!(plan.contains(&step.target("B").name("dist-mingw")));
assert!(plan.contains(&step.target("B").name("dist-rustc")));
assert!(plan.contains(&step.target("B").name("dist-std")));
assert!(!plan.contains(&step.target("B").name("dist-src")));
}
#[test]
fn dist_with_targets_and_hosts() {
let build = build(&["dist"], &["B"], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(!plan.iter().any(|s| s.host == "B"));
assert!(!plan.iter().any(|s| s.host == "C"));
assert!(plan.contains(&step.name("dist-docs")));
assert!(plan.contains(&step.name("dist-mingw")));
assert!(plan.contains(&step.name("dist-rustc")));
assert!(plan.contains(&step.name("dist-std")));
assert!(plan.contains(&step.name("dist-src")));
assert!(plan.contains(&step.target("B").name("dist-docs")));
assert!(plan.contains(&step.target("B").name("dist-mingw")));
assert!(plan.contains(&step.target("B").name("dist-rustc")));
assert!(plan.contains(&step.target("B").name("dist-std")));
assert!(!plan.contains(&step.target("B").name("dist-src")));
assert!(plan.contains(&step.target("C").name("dist-docs")));
assert!(plan.contains(&step.target("C").name("dist-mingw")));
assert!(!plan.contains(&step.target("C").name("dist-rustc")));
assert!(plan.contains(&step.target("C").name("dist-std")));
assert!(!plan.contains(&step.target("C").name("dist-src")));
}
#[test]
fn dist_target_with_target_flag() {
let build = build(&["dist", "--target=C"], &["B"], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(!plan.iter().any(|s| s.target == "A"));
assert!(!plan.iter().any(|s| s.target == "B"));
assert!(!plan.iter().any(|s| s.host == "B"));
assert!(!plan.iter().any(|s| s.host == "C"));
assert!(plan.contains(&step.target("C").name("dist-docs")));
assert!(plan.contains(&step.target("C").name("dist-mingw")));
assert!(!plan.contains(&step.target("C").name("dist-rustc")));
assert!(plan.contains(&step.target("C").name("dist-std")));
assert!(!plan.contains(&step.target("C").name("dist-src")));
}
#[test]
fn dist_host_with_target_flag() {
let build = build(&["dist", "--host=B", "--target=B"], &["B"], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
assert!(!plan.iter().any(|s| s.target == "A"));
assert!(!plan.iter().any(|s| s.target == "C"));
assert!(!plan.iter().any(|s| s.host == "B"));
assert!(!plan.iter().any(|s| s.host == "C"));
assert!(plan.contains(&step.target("B").name("dist-docs")));
assert!(plan.contains(&step.target("B").name("dist-mingw")));
assert!(plan.contains(&step.target("B").name("dist-rustc")));
assert!(plan.contains(&step.target("B").name("dist-std")));
assert!(plan.contains(&step.target("B").name("dist-src")));
let all = rules.expand(&plan);
println!("all rules: {:#?}", all);
assert!(!all.contains(&step.name("rustc")));
rustbuild: Add support for compiling Cargo This commit adds support to rustbuild for compiling Cargo as part of the release process. Previously rustbuild would simply download a Cargo snapshot and repackage it. With this change we should be able to turn off artifacts from the rust-lang/cargo repository and purely rely on the artifacts Cargo produces here. The infrastructure added here is intended to be extensible to other components, such as the RLS. It won't exactly be a one-line addition, but the addition of Cargo didn't require too much hooplah anyway. The process for release Cargo will now look like: * The rust-lang/rust repository has a Cargo submodule which is used to build a Cargo to pair with the rust-lang/rust release * Periodically we'll update the cargo submodule as necessary on rust-lang/rust's master branch * When branching beta we'll create a new branch of Cargo (as we do today), and the first commit to the beta branch will be to update the Cargo submodule to this exact revision. * When branching stable, we'll ensure that the Cargo submodule is updated and then make a stable release. Backports to Cargo will look like: * Send a PR to cargo's master branch * Send a PR to cargo's release branch (e.g. rust-1.16.0) * Send a PR to rust-lang/rust's beta branch updating the submodule * Eventually send a PR to rust-lang/rust's master branch updating the submodule For reference, the process to add a new component to the rust-lang/rust release would look like: * Add `$foo` as a submodule in `src/tools` * Add a `tool-$foo` step which compiles `$foo` with the specified compiler, likely mirroring what Cargo does. * Add a `dist-$foo` step which uses `src/tools/$foo` and the `tool-$foo` output to create a rust-installer package for `$foo` likely mirroring what Cargo does. * Update the `dist-extended` step with a new dependency on `dist-$foo` * Update `src/tools/build-manifest` for the new component.
2017-02-16 00:57:06 +01:00
assert!(!all.contains(&step.name("build-crate-test").stage(1)));
// all stage0 compiles should be for the build target, A
for step in all.iter().filter(|s| s.stage == 0) {
if !step.name.contains("build-crate") {
continue
}
println!("step: {:?}", step);
assert!(step.host != "B");
assert!(step.target != "B");
assert!(step.host != "C");
assert!(step.target != "C");
}
}
#[test]
fn build_default() {
let build = build(&["build"], &["B"], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
let step = super::Step {
name: "",
stage: 2,
host: &build.config.build,
target: &build.config.build,
};
// rustc built for all for of (A, B) x (A, B)
assert!(plan.contains(&step.name("librustc")));
assert!(plan.contains(&step.target("B").name("librustc")));
assert!(plan.contains(&step.host("B").target("A").name("librustc")));
assert!(plan.contains(&step.host("B").target("B").name("librustc")));
// rustc never built for C
assert!(!plan.iter().any(|s| {
s.name.contains("rustc") && (s.host == "C" || s.target == "C")
}));
// test built for everything
assert!(plan.contains(&step.name("libtest")));
assert!(plan.contains(&step.target("B").name("libtest")));
assert!(plan.contains(&step.host("B").target("A").name("libtest")));
assert!(plan.contains(&step.host("B").target("B").name("libtest")));
assert!(plan.contains(&step.host("A").target("C").name("libtest")));
assert!(plan.contains(&step.host("B").target("C").name("libtest")));
let all = rules.expand(&plan);
println!("all rules: {:#?}", all);
assert!(all.contains(&step.name("rustc")));
assert!(all.contains(&step.name("libstd")));
}
#[test]
fn build_filtered() {
let build = build(&["build", "--target=C"], &["B"], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
assert!(!plan.iter().any(|s| s.name.contains("rustc")));
assert!(plan.iter().all(|s| {
!s.name.contains("test") || s.target == "C"
}));
}
#[test]
fn test_default() {
let build = build(&["test"], &[], &[]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
assert!(plan.iter().all(|s| s.host == "A"));
assert!(plan.iter().all(|s| s.target == "A"));
assert!(plan.iter().any(|s| s.name.contains("-ui")));
assert!(plan.iter().any(|s| s.name.contains("cfail")));
assert!(plan.iter().any(|s| s.name.contains("cfail-full")));
assert!(plan.iter().any(|s| s.name.contains("codegen-units")));
assert!(plan.iter().any(|s| s.name.contains("debuginfo")));
assert!(plan.iter().any(|s| s.name.contains("docs")));
assert!(plan.iter().any(|s| s.name.contains("error-index")));
assert!(plan.iter().any(|s| s.name.contains("incremental")));
assert!(plan.iter().any(|s| s.name.contains("linkchecker")));
assert!(plan.iter().any(|s| s.name.contains("mir-opt")));
assert!(plan.iter().any(|s| s.name.contains("pfail")));
assert!(plan.iter().any(|s| s.name.contains("rfail")));
assert!(plan.iter().any(|s| s.name.contains("rfail-full")));
assert!(plan.iter().any(|s| s.name.contains("rmake")));
assert!(plan.iter().any(|s| s.name.contains("rpass")));
assert!(plan.iter().any(|s| s.name.contains("rpass-full")));
assert!(plan.iter().any(|s| s.name.contains("rustc-all")));
assert!(plan.iter().any(|s| s.name.contains("rustdoc")));
assert!(plan.iter().any(|s| s.name.contains("std-all")));
assert!(plan.iter().any(|s| s.name.contains("test-all")));
assert!(plan.iter().any(|s| s.name.contains("tidy")));
assert!(plan.iter().any(|s| s.name.contains("valgrind")));
}
#[test]
fn test_with_a_target() {
let build = build(&["test", "--target=C"], &[], &["C"]);
let rules = super::build_rules(&build);
let plan = rules.plan();
println!("rules: {:#?}", plan);
assert!(plan.iter().all(|s| s.stage == 2));
assert!(plan.iter().all(|s| s.host == "A"));
assert!(plan.iter().all(|s| s.target == "C"));
assert!(plan.iter().any(|s| s.name.contains("-ui")));
assert!(!plan.iter().any(|s| s.name.contains("ui-full")));
assert!(plan.iter().any(|s| s.name.contains("cfail")));
assert!(!plan.iter().any(|s| s.name.contains("cfail-full")));
assert!(plan.iter().any(|s| s.name.contains("codegen-units")));
assert!(plan.iter().any(|s| s.name.contains("debuginfo")));
assert!(!plan.iter().any(|s| s.name.contains("docs")));
assert!(!plan.iter().any(|s| s.name.contains("error-index")));
assert!(plan.iter().any(|s| s.name.contains("incremental")));
assert!(!plan.iter().any(|s| s.name.contains("linkchecker")));
assert!(plan.iter().any(|s| s.name.contains("mir-opt")));
assert!(plan.iter().any(|s| s.name.contains("pfail")));
assert!(plan.iter().any(|s| s.name.contains("rfail")));
assert!(!plan.iter().any(|s| s.name.contains("rfail-full")));
assert!(!plan.iter().any(|s| s.name.contains("rmake")));
assert!(plan.iter().any(|s| s.name.contains("rpass")));
assert!(!plan.iter().any(|s| s.name.contains("rpass-full")));
assert!(!plan.iter().any(|s| s.name.contains("rustc-all")));
assert!(!plan.iter().any(|s| s.name.contains("rustdoc")));
assert!(plan.iter().any(|s| s.name.contains("std-all")));
assert!(plan.iter().any(|s| s.name.contains("test-all")));
assert!(!plan.iter().any(|s| s.name.contains("tidy")));
assert!(plan.iter().any(|s| s.name.contains("valgrind")));
}
}