Ensure libraries built in stage0 have unique metadata
Issue #50786 shows a case with local rebuild where the libraries built
by stage0 had the same suffix as stage0's own, and were accidentally
loaded by that stage0 rustc when compiling `librustc_trans`.
Now we set `__CARGO_DEFAULT_LIB_METADATA` to "bootstrap" during stage0,
rather than the release channel like usual, so the library suffix will
always be completely distinct from the stage0 compiler.
Remove LazyBTreeMap.
It was introduced in #50240 to avoid an allocation when creating a new
BTreeMap, which gave some speed-ups. But then #50352 made that the
default behaviour for BTreeMap, so LazyBTreeMap is no longer necessary.
Fix rustdoc panic with `impl Trait` in type parameters
Fixes#50702.
I'm not sure `impl Trait`s neither in arguments nor in return types are supposed to work, though.
read2: Use inner function instead of closure
Very minor thing, but there doesn't appear to be a reason to use a closure here.
Generated code is identical in my tests, but I believe it's clearer that nothing from the environment is being used.
Check for empty `.stderr` and `.stdout` files in UI test directories.
Empty files could still pass testing for `compile-pass` tests with no output
so they can get into the repo accidentally, but they are not necessary and can
be removed.
Reenable the MergeFunctions pass
The crash that happened in #23566 doesn't happen anymore with the LLVM mergefunc
pass enabled and it hugely reduces code size (for example it shaves off 10% of the
final Servo executable). This patch reenables it.
For those wondering, [here are the docs from LLVM about this pass](http://llvm.org/docs/MergeFunctions.html).
rustdoc: deprecate `#![doc(passes, plugins, no_default_passes)]`
Closes https://github.com/rust-lang/rust/issues/48164
Blocked on https://github.com/rust-lang/rust/pull/50541 - this includes those changes, which were necessary to create the UI test
cc https://github.com/rust-lang/rust/issues/44136
Turns out, there were special attributes to mess with rustdoc passes and plugins! Who knew! Since we deprecated the CLI flags for this functionality, it makes sense that we do the same for the attributes.
This PR also introduces a `#![doc(document_private_items)]` attribute, to match the `--document-private-items` flag introduced in https://github.com/rust-lang/rust/pull/44138 when the passes/plugins flags were deprecated.
I haven't done a search to see whether these attributes are being used at all, but if the flags were any indication, i don't expect to see any users of these.
Fix `fn main() -> impl Trait` for non-`Termination` trait
Fixes#50595.
This bug currently affects stable. Why I think we can go for hard error:
- It will in stable for at most one cycle and there is no legitimate reason to abuse it, nor any known uses in the wild.
- It only affects `bin` crates (which have a `main`), so there is little practical difference between a hard error or a deny lint, both are a one line fix.
The fix was to just unshadow a variable. Thanks @nikomatsakis for the mentoring!
r? @nikomatsakis
Don't unconditionally set CLOEXEC twice on every fd we open on Linux
Previously, every `open64` was accompanied by a `ioctl(…, FIOCLEX)`,
because some old Linux version would ignore the `O_CLOEXEC` flag we pass
to the `open64` function.
Now, we check whether the `CLOEXEC` flag is set on the first file we
open – if it is, we won't do extra syscalls for every opened file. If it
is not set, we fall back to the old behavior of unconditionally calling
`ioctl(…, FIOCLEX)` on newly opened files.
On old Linuxes, this amounts to one extra syscall per process, namely
the `fcntl(…, F_GETFD)` call to check the `CLOEXEC` flag.
On new Linuxes, this reduces the number of syscalls per opened file by
one, except for the first file, where it does the same number of
syscalls as before (`fcntl(…, F_GETFD)` to check the flag instead of
`ioctl(…, FIOCLEX)` to set it).
Review proc macro API 1.2
cc https://github.com/rust-lang/rust/issues/38356
Summary of applied changes:
- Documentation for proc macro API 1.2 is expanded.
- Renamed APIs: `Term` -> `Ident`, `TokenTree::Term` -> `TokenTree::Ident`, `Op` -> `Punct`, `TokenTree::Op` -> `TokenTree::Punct`, `Op::op` -> `Punct::as_char`.
- Removed APIs: `Ident::as_str`, use `Display` impl for `Ident` instead.
- New APIs (not stabilized in 1.2): `Ident::new_raw` for creating a raw identifier (I'm not sure `new_x` it's a very idiomatic name though).
- Runtime changes:
- `Punct::new` now ensures that the input `char` is a valid punctuation character in Rust.
- `Ident::new` ensures that the input `str` is a valid identifier in Rust.
- Lifetimes in proc macros are now represented as two joint tokens - `Punct('\'', Spacing::Joint)` and `Ident("lifetime_name_without_quote")` similarly to multi-character operators.
- Stabilized APIs: None yet.
A bit of motivation for renaming (although it was already stated in the review comments):
- With my compiler frontend glasses on `Ident` is the single most appropriate name for this thing, *especially* if we are doing input validation on construction. `TokenTree::Ident` effectively wraps `token::Ident` or `ast::Ident + is_raw`, its meaning is "identifier" and it's already named `ident` in declarative macros.
- Regarding `Punct`, the motivation is that `Op` is actively misleading. The thing doesn't mean an operator, it's neither a subset of operators (there is non-operator punctuation in the language), nor superset (operators can be multicharacter while this thing is always a single character). So I named it `Punct` (first proposed in [the original RFC](https://github.com/rust-lang/rfcs/pull/1566), then [by @SimonSapin](https://github.com/rust-lang/rust/issues/38356#issuecomment-276676526)) , together with input validation it's now a subset of ASCII punctuation character category (`u8::is_ascii_punctuation`).
Implement RFC 2056 trivial constraints in where clauses
This is an implementation of the new behaviour for #48214. Tests are mostly updated to show the effects of this. Feature gate hasn't been added yet.
Some things that are worth noting and are maybe not want we want
* `&mut T: Copy` doesn't allow as much as someone might expect because there is often an implicit reborrow.
* ~There isn't a check that a where clause is well-formed any more, so `where Vec<str>: Debug` is now allowed (without a `str: Sized` bound).~
r? @nikomatsakis
Remove ScopeTarget and LoopIdResult
* Remove ScopeTarget in preparation of label-break-value (PR #50045)
* Replace LoopIdResult by Result which is possible now thanks to commit 8ac65af81f " Implement Encodable and Decodable for Result."
r? @eddyb
rustdoc: replace most (e)println! statements with structured warnings/errors
Turns out, the rustc diagnostic handler doesn't need a whole lot of setup that we weren't already doing. For errors that occur outside a "dealing with source code" context, we can just use the format/color config we were already parsing and make up a `Handler` that we can emit structured warnings/errors from. So i did that. This will make it way easier to test things with `rustdoc-ui` tests, since those require the JSON error output. (In fact, this PR is a yak shave for a different one where i was trying to do just that. `>_>`)
The grammar defines identifiers in terms of XID_start and XID_continue,
but this is referring to the unstable non_ascii_idents feature.
The documentation implies that non_ascii_idents is forthcoming, but this
is left over from pre-1.0 documentation; in reality, non_ascii_idents
has been without even an RFC for several years now, and will not be
stabilized anytime soon. Furthermore, according to the tracking issue at
https://github.com/rust-lang/rust/issues/28979 , it's highly
questionable whether or not this feature will use XID_start or
XID_continue even when or if non_ascii_idents is stabilized.
This commit fixes this by respecifying identifiers as the usual
[a-zA-Z_][a-zA-Z0-9_]*
Issue #50786 shows a case with local rebuild where the libraries built
by stage0 had the same suffix as stage0's own, and were accidentally
loaded by that stage0 rustc when compiling `librustc_trans`.
Now we set `__CARGO_DEFAULT_LIB_METADATA` to "bootstrap" during stage0,
rather than the release channel like usual, so the library suffix will
always be completely distinct from the stage0 compiler.
