rework the README.md for rustc and add other readmes
OK, so, long ago I committed to the idea of trying to write some high-level documentation for rustc. This has proved to be much harder for me to get done than I thought it would! This PR is far from as complete as I had hoped, but I wanted to open it so that people can give me feedback on the conventions that it establishes. If this seems like a good way forward, we can land it and I will open an issue with a good check-list of things to write (and try to take down some of them myself).
Here are the conventions I established on which I would like feedback.
**Use README.md files**. First off, I'm aiming to keep most of the high-level docs in `README.md` files, rather than entries on forge. My thought is that such files are (a) more discoverable than forge and (b) closer to the code, and hence can be edited in a single PR. However, since they are not *in the code*, they will naturally get out of date, so the intention is to focus on the highest-level details, which are least likely to bitrot. I've included a few examples of common functions and so forth, but never tried to (e.g.) exhaustively list the names of functions and so forth.
- I would like to use the tidy scripts to try and check that these do not go out of date. Future work.
**librustc/README.md as the main entrypoint.** This seems like the most natural place people will look first. It lays out how the crates are structured and **is intended** to give pointers to the main data structures of the compiler (I didn't update that yet; the existing material is terribly dated).
**A glossary listing abbreviations and things.** It's much harder to read code if you don't know what some obscure set of letters like `infcx` stands for.
**Major modules each have their own README.md that documents the high-level idea.** For example, I wrote some stuff about `hir` and `ty`. Both of them have many missing topics, but I think that is roughly the level of depth that would be good. The idea is to give people a "feeling" for what the code does.
What is missing primarily here is lots of content. =) Here are some things I'd like to see:
- A description of what a QUERY is and how to define one
- Some comments for `librustc/ty/maps.rs`
- An overview of how compilation proceeds now (i.e., the hybrid demand-driven and forward model) and how we would like to see it going in the future (all demand-driven)
- Some coverage of how incremental will work under red-green
- An updated list of the major IRs in use of the compiler (AST, HIR, TypeckTables, MIR) and major bits of interesting code (typeck, borrowck, etc)
- More advice on how to use `x.py`, or at least pointers to that
- Good choice for `config.toml`
- How to use `RUST_LOG` and other debugging flags (e.g., `-Zverbose`, `-Ztreat-err-as-bug`)
- Helpful conventions for `debug!` statement formatting
cc @rust-lang/compiler @mgattozzi
rustc: Forbid interpolated tokens in the HIR
Right now the HIR contains raw `syntax::ast::Attribute` structure but nowadays
these can contain arbitrary tokens. One variant of the `Token` enum is an
"interpolated" token which basically means to shove all the tokens for a
nonterminal in this position. A "nonterminal" in this case is roughly analagous
to a macro argument:
macro_rules! foo {
($a:expr) => {
// $a is a nonterminal as an expression
}
}
Currently nonterminals contain namely items and expressions, and this poses a
problem for incremental compilation! With incremental we want a stable hash of
all HIR items, but this means we may transitively need a stable hash *of the
entire AST*, which is certainly not stable w/ node ids and whatnot. Hence today
there's a "bug" where the "stable hash" of an AST is just the raw hash value of
the AST, and this only arises with interpolated nonterminals. The downside of
this approach, however, is that a bunch of errors get spewed out during
compilation about how this isn't a great idea.
This PR is focused at fixing these warnings, basically deleting them from the
compiler. The implementation here is to alter attributes as they're lowered from
the AST to HIR, expanding all nonterminals in-place as we see them. This code
for expanding a nonterminal to a token stream already exists for the
`proc_macro` crate, so we basically just reuse the same implementation there.
After this PR it's considered a bug to have an `Interpolated` token and hence
the stable hash implementation simply uses `bug!` in this location.
Closes#40946
hide internal types/traits from std docs via new #[doc(masked)] attribute
Fixes#43701 (hopefully for good this time)
This PR introduces a new parameter to the `#[doc]` attribute that rustdoc looks for on `extern crate` statements. When it sees `#[doc(masked)]` on such a statement, it hides traits and types from that crate from appearing in either the "Trait Implementations" section of many type pages, or the "Implementors" section of trait pages. This is then applied to the `libc`/`rand`/`compiler_builtins` imports in libstd to prevent those crates from creating broken links in the std docs.
Like in #43348, this also introduces a feature gate, `doc_masked`, that controls the use of this parameter.
To view the std docs generated with this change, head to https://tonberry.quietmisdreavus.net/std-43701/std/index.html.
Right now the HIR contains raw `syntax::ast::Attribute` structure but nowadays
these can contain arbitrary tokens. One variant of the `Token` enum is an
"interpolated" token which basically means to shove all the tokens for a
nonterminal in this position. A "nonterminal" in this case is roughly analagous
to a macro argument:
macro_rules! foo {
($a:expr) => {
// $a is a nonterminal as an expression
}
}
Currently nonterminals contain namely items and expressions, and this poses a
problem for incremental compilation! With incremental we want a stable hash of
all HIR items, but this means we may transitively need a stable hash *of the
entire AST*, which is certainly not stable w/ node ids and whatnot. Hence today
there's a "bug" where the "stable hash" of an AST is just the raw hash value of
the AST, and this only arises with interpolated nonterminals. The downside of
this approach, however, is that a bunch of errors get spewed out during
compilation about how this isn't a great idea.
This PR is focused at fixing these warnings, basically deleting them from the
compiler. The implementation here is to alter attributes as they're lowered from
the AST to HIR, expanding all nonterminals in-place as we see them. This code
for expanding a nonterminal to a token stream already exists for the
`proc_macro` crate, so we basically just reuse the same implementation there.
After this PR it's considered a bug to have an `Interpolated` token and hence
the stable hash implementation simply uses `bug!` in this location.
Closes#40946
rustbuild: with --no-fail-fast, report the specific commands that failed
I'm not sure this is the most elegant way of doing it, I'm still a bit of a rust noob. I tried `Vec<Command>` and keeping `Cell` instead of `RefCell` but couldn't fight my way past the borrow errors, this was the first arrangement that I could make work.
Replace str's transmute() calls with pointer casts
After the following conversation in #rust-lang:
```
[14:43:50] <Ixrec> TIL the implementation of from_utf_unchecked is literally just "mem::transmute(x)"
[14:43:59] <Ixrec> no wonder people keep saying transmute is overpowered
[15:15:30] <eddyb> Ixrec: it should be a pointer cast lol
[15:15:46] <eddyb> unless it doesn't let you
[16:50:34] <Ixrec> https://play.rust-lang.org/?gist=d1e6b629ad9ec1baf64ce261c63845e6&version=stable seems like it does let me
[16:52:35] <eddyb> Ixrec: yeah that's the preferred impl
[16:52:46] <eddyb> Ixrec: it just wasn't in 1.0
[16:52:50] <eddyb> IIRC
[16:53:00] <eddyb> (something something fat pointers)
```
Since I already wrote half of the preferred impls in the playground, might as well make an actual PR.
Document thread builder panics for nul bytes in thread names
This seems to have been undocumented. Mention this where the name is set
(Builder::name) and where the panic could happen (Builder::spawn).
Thread::new is private and I think the builder is the only user where
this matters. A short comment was added to "document" Thread::new too.
Add proper help line for `-C inline threshold`
Looks like someone accidentally some words when adding this.
This also remove a period on a different help line for consistency, as no options have a period.
Add Cow<str> -> Box<Error> impls.
Considering how impls exist for `String` and `&str`, it makes sense to also add an impl for `Cow<str>` as well.
This would allow converting `String::from_utf8_lossy` directly into a `Box<Error>` or `io::Error` without having to add an extra `into_ownd()`.
incr.comp.: Compute fingerprint for all query results.
This PR enables query result fingerprinting in incremental mode. This is an essential piece of infrastructure for red/green tracking. We don't do anything with the fingerprints yet but merging the infrastructure should protect it from bit-rotting and will make it easier to start measuring its performance impact (and thus let us determine if we should switch to a faster hashing algorithm rather sooner than later).
Note, this PR also includes the changes from https://github.com/rust-lang/rust/pull/43887 which I'm therefore closing. No need to re-review the first commit though.
r? @nikomatsakis
- Don't hash traits in scope as part of HIR hashing any more.
- Some queries returned DefIndexes from other crates.
- Provide a generic way of stably hashing maps (not used everywhere yet).
This makes sure that we don't introduce strange cases where we have
nodes outside the query system that could break red/green tracking
and it will allow to keep red/green neatly encapsulated within the
DepGraph implementation.
Run the miri test suite on the aux builder and travis
Reopen of #38350
see https://github.com/rust-lang/rust/pull/43340#issuecomment-316940762 for earlier discussion
Rationale for running miri's test suite in rustc's CI is that miri currently contains many features that we want in const eval in the future, and these features would break if the test suite is not run.
fixes#44077
r? @nikomatsakis
cc @eddyb
Refactor translation unit partitioning/collection as a query
This commit is targeted at #44486 with the ultimate goal of making the `collect_and_partition_translation_items` function a query. This mostly just involved query-ifying a few other systems along with plumbing the tcx instead of `SharedCrateContext` in a few locations.
Currently this only tackles the first bullet of #44486 and doesn't add a dedicated query for a particular codegen unit. I wasn't quite sure how to do that yet but figured this was good to put up.
Closes#44486
