DST coercions and DST fields in structs
The commits are not quite stand alone, I should probably squash them together before landing. In particular if you review the individual commits, then you'll see some scrappy stuff that gets fixed in later commits. But reading the commits in order might be easier to get an overall idea of what is going on.
The first commit includes putting back time zone into our time library - @pcwalton removed that as part of his de-~str'ing, but I had already converted it to use StrBuf, so we may as well leave it in. Update: no longer, this is removed in a later commit.
[breaking-change]
1. The internal layout for traits has changed from (vtable, data) to (data, vtable). If you were relying on this in unsafe transmutes, you might get some very weird and apparently unrelated errors. You should not be doing this! Prefer not to do this at all, but if you must, you should use raw::TraitObject rather than hardcoding rustc's internal representation into your code.
2. The minimal type of reference-to-vec-literals (e.g., `&[1, 2, 3]`) is now a fixed size vec (e.g., `&[int, ..3]`) where it used to be an unsized vec (e.g., `&[int]`). If you want the unszied type, you must explicitly give the type (e.g., `let x: &[_] = &[1, 2, 3]`). Note in particular where multiple blocks must have the same type (e.g., if and else clauses, vec elements), the compiler will not coerce to the unsized type without a hint. E.g., `[&[1], &[1, 2]]` used to be a valid expression of type '[&[int]]'. It no longer type checks since the first element now has type `&[int, ..1]` and the second has type &[int, ..2]` which are incompatible.
3. The type of blocks (including functions) must be coercible to the expected type (used to be a subtype). Mostly this makes things more flexible and not less (in particular, in the case of coercing function bodies to the return type). However, in some rare cases, this is less flexible. TBH, I'm not exactly sure of the exact effects. I think the change causes us to resolve inferred type variables slightly earlier which might make us slightly more restrictive. Possibly it only affects blocks with unreachable code. E.g., `if ... { fail!(); "Hello" }` used to type check, it no longer does. The fix is to add a semicolon after the string.
The privacy pass of the compiler was previously not taking into account the
privacy of foreign items, or bindings to external functions. This commit fixes
this oversight by encoding the visibility of foreign items into the metadata for
each crate.
Any code relying on this will start to fail to compile and the bindings must be
marked with `pub` to indicate that they can be used externally.
Closes#16725
[breaking-change]
If a task is spinning in an accept loop, there is currently no method of gracefully shutting it down. This PR introduces a way to do so by cloning the acceptor and implementing a close_accept method to unblocking any pending acceptor.
As with other I/O methods like this, it is `#[experimental]` from the start and sadly carries with it a good deal of code to support it. Much of the complication is from the fact that you can now concurrently accept on the same socket.
I tried to add a good deal of tests for this change, but another set of eyes is always appreciated!
This commits takes a similar strategy to the previous commit to implement
close_accept and clone for the native win32 pipes implementation.
Closes#15595
This commits implements {Tcp,Unix}Acceptor::{clone,close_accept} methods for
all of librustuv.
This implementation rewrites much of Access, AccessTimeout, and AcceptTimeout to
have type parameter for shared state that all acceptors share (a shared queue of
sockets). The incoming/outgoing channels have been removed as all timeouts and
such are now managed on the event loop rather than concurrently.
This commits implements {Tcp,Unix}Acceptor::{clone,close_accept} methods for
unix. A windows implementation is coming in a later commit.
The clone implementation is based on atomic reference counting (as with all
other clones), and the close_accept implementation is based on selecting on a
self-pipe which signals that a close has been seen.
Current version of rust fails when casting from bool, e.g.
```rust
fn main() {
let _a = false as uint;
let _b = true as uint;
let _c: [bool, ..false as uint];
let _d: [bool, ..true as uint];
// _a and _b work, but _c and _d result in an error
// error: expected constant expr for vector length: can't cast str to uint
}
```
This commit makes it work as expected.
Use ExactSize::len() and defer to its decisions about overly defensive
assertions. Remove the length double-check and simply put a failure
case if the Zip finds an uneven end in .next_back().
Fixing this up since I think I wrote this, and it's been known to
confuse rusties (PR #15886).
Use ExactSize::len() and defer to its decisions about overly defensive
assertions. Remove the length double-check and simply put a failure
case if the Zip finds an uneven end in .next_back().
Fixing this up since I think I wrote this, and it's been known to
confuse rusties (PR#15886).
Also:
* Remove unseeming repetition.
* By now, the reader has already heard that Rust is safe by default, so
reduce the overlong sentence, making it easier to read.
These are somewhat stop-gap solutions to address #16625
core: Separate failure formatting in str methods slice, slice_to, slice_from
Use a separate inline-never function to format failure message for
str::slice() errors.
Using strcat's idea, this makes sure no formatting code from failure is
inlined when str::slice() is inlined. The number of `unreachable` being
inlined when usingi `.slice()` drops from 5 to just 1.
The testcase:
```
#![crate_type = "lib"]
pub fn slice(x: &str, a: uint, b: uint) -> &str {
x.slice(a, b)
}
```
shrinks from 16.9 kB to 3.3 kB llvm IR, and the number of `unreachable` drops from 5 to 1.
Also:
* Remove unseeming repetition.
* By now, the reader has already heard that Rust is safe by default, so
reduce the overlong sentence, making it easier to read.
I was doing a lot of parsing ascii strings, and the generic bsearch functions in `tables.rs` came up very high in the profile.
This should avoid calling those functions for simple ASCII range chars.
There is a check in TwoWaySearcher::new to determine whether the needle is periodic. This is needed because during searching when a match fails, we cannot advance the position by the entire length of the needle when it is periodic, but can only advance by the length of the period.
The reason "bananas".contains("nana") (and similar searches) were returning false was because the periodicity check was wrong.
Closes#16589
Also, thanks to @Gankro, who came up with many buggy examples.
