Rollup merge of #63400 - petrochenkov:resplit, r=eddyb
Try to break resolve into more isolated parts
Some small step towards resolve librarification.
"Late resolution" is the pass that resolves most of names in a crate beside imports and macros.
It runs when the crate is fully expanded and its module structure is fully built.
So we just walk through the crate and resolve all the expressions, types, etc.
This pass is pretty self-contained, but it was previously done by implementing `Visitor` on the whole `Resolver` (which is used for many other tasks), and fields specific to this pass were indiscernible from the global `Resolver` state.
This PR moves the late resolution pass into a separate visitor and a separate file, fields specific to this visitor are moved from `Resolver` as well.
I'm especially happy about `current_module` being removed from `Resolver`.
It was used even for operations not related to visiting and changing the `current_module` position in process.
It was also used as an implicit argument for some functions used in this style
```rust
let orig_current_module = mem::replace(&mut self.current_module, module);
self.resolve_ident_somewhere();
self.current_module = orig_current_module;
```
and having effects on e.g. privacy checking somewhere deeply inside `resolve_ident_somewhere`.
Now we explicitly pass a `ParentScope` to those functions instead, which includes the module and some other data describing our position in the crate relatively to which we resolve names.
Rustdoc was one of the users of `current_module`, it set it for resolving intra-doc links.
Now it passes it explicitly as an argument as well (I also supported resolving paths from rustdoc in unnamed blocks as a drive-by fix).
Visibility resolution is also changed to use early resolution (which is correct because it's used during the work of `BuildReducedGraphVisitor`, i.e. integration of a new AST fragment into the existing partially built module structures.) instead of untimely late resolution (which worked only due to restrictions on paths in visibilities like inability to refer to anything except ancestor modules).
This slightly regresses its diagnostics because late resolution has a more systematic error detection and recovery currently.
Due to changes in `current_module` and visibilities `BuildReducedGraphVisitor` ended up almost as heavily affected by this refactoring as late resolution.
Fixes https://github.com/rust-lang/rust/issues/63223 (due to visibility resolution changes).
Fix calls to resolver from rustdoc and HIR lowering
Cleanup some surrounding code.
Support resolution of intra doc links in unnamed block scopes.
(Paths from rustdoc now use early resolution and no longer need results of late resolution like all the built ribs.)
Move methods logically belonging to build-reduced-graph into `impl BuildReducedGraphVisitor` and `build_reduced_graph.rs`
Move types mostly specific to late resolution closer to the late resolution visitor
resolve: Move late resolution into a separate visitor
Move `Resolver` fields specific to late resolution to the new visitor.
The `current_module` field from `Resolver` is replaced with two `current_module`s in `LateResolutionVisitor` and `BuildReducedGraphVisitor`.
Outside of those visitors `current_module` is replaced by passing `parent_scope` to more functions and using the parent module from it.
Visibility resolution no longer have access to later resolution methods and has to use early resolution, so its diagnostics in case of errors regress slightly.
bors [Sat, 10 Aug 2019 06:14:13 +0000 (06:14 +0000)]
Auto merge of #63428 - Centril:rollup-c2ru1z1, r=Centril
Rollup of 7 pull requests
Successful merges:
- #63056 (Give built-in macros stable addresses in the standard library)
- #63337 (Tweak mismatched types error)
- #63350 (Use associated_type_bounds where applicable - closes #61738)
- #63394 (Add test for issue 36804)
- #63399 (More explicit diagnostic when using a `vec![]` in a pattern)
- #63419 (check against more collisions for TypeId of fn pointer)
- #63423 (Mention that tuple structs are private if any of their fields are)
Rollup merge of #63399 - estebank:vec-in-pat, r=Centril
More explicit diagnostic when using a `vec![]` in a pattern
```
error: unexpected `(` after qualified path
--> $DIR/vec-macro-in-pattern.rs:3:14
|
LL | Some(vec![x]) => (),
| ^^^^^^^
| |
| unexpected `(` after qualified path
| in this macro invocation
| use a slice pattern here instead
|
= help: for more information, see https://doc.rust-lang.org/edition-guide/rust-2018/slice-patterns.html
= note: this warning originates in a macro outside of the current crate (in Nightly builds, run with -Z external-macro-backtrace for more info)
```
Rollup merge of #63337 - estebank:break-ee0308, r=Centril
Tweak mismatched types error
- Change expected/found for type mismatches in `break`
- Be more accurate when talking about diverging match arms
- Tweak wording of function without a return value
- Suggest calling bare functions when their return value can be coerced to the expected type
- Give more parsing errors when encountering `foo(_, _, _)`
Fix #51767, fix #62677, fix #63136, cc #37384, cc #35241, cc #51669.
Rollup merge of #63056 - petrochenkov:macstd2, r=alexcrichton
Give built-in macros stable addresses in the standard library
Continuation of https://github.com/rust-lang/rust/pull/62086.
Derive macros corresponding to traits from libcore are now available through the same paths as those traits:
- `Clone` - `{core,std}::clone::Clone`
- `PartialEq` - `{core,std}::cmp::PartialEq`
- `Eq` - `{core,std}::cmp::Eq`
- `PartialOrd` - `{core,std}::cmp::PartialOrd`
- `Ord` - `{core,std}::cmp::Ord`
- `Default` - `{core,std}::default::Default`
- `Debug` - `{core,std}::fmt::Debug`
- `Hash` - `{core,std}::hash::Hash`
- `Copy` - `{core,std}::marker::Copy`
Fn-like built-in macros are now available through libcore and libstd's root module, by analogy with non-builtin macros defined by libcore and libstd:
```rust
{core,std}::{
__rust_unstable_column,
asm,
assert,
cfg,
column,
compile_error,
concat,
concat_idents,
env,
file,
format_args,
format_args_nl,
global_asm,
include,
include_bytes,
include_str,
line,
log_syntax,
module_path,
option_env,
stringify,
trace_macros,
}
```
Derive macros without a corresponding trait in libcore or libstd are still available only through prelude (also see https://github.com/rust-lang/rust/pull/62507).
Attribute macros also keep being available only through prelude, mostly because they don't have an existing practice to follow. An advice from the library team on their eventual placement would be appreciated.
```rust
RustcDecodable,
RustcEncodable,
bench,
global_allocator,
test,
test_case,
```
Esteban Küber [Fri, 9 Aug 2019 01:24:00 +0000 (18:24 -0700)]
More explicit diagnostic when using a `vec![]` in a pattern
```
error: unexpected `(` after qualified path
--> $DIR/vec-macro-in-pattern.rs:3:14
|
LL | Some(vec![x]) => (),
| ^^^^^^^
| |
| unexpected `(` after qualified path
| in this macro invocation
| use a slice pattern here instead
|
= help: for more information, see https://doc.rust-lang.org/edition-guide/rust-2018/slice-patterns.html
= note: this warning originates in a macro outside of the current crate (in Nightly builds, run with -Z external-macro-backtrace for more info)
```
bors [Fri, 9 Aug 2019 12:14:57 +0000 (12:14 +0000)]
Auto merge of #63408 - Centril:rollup-skqrez3, r=Centril
Rollup of 7 pull requests
Successful merges:
- #62672 (Deprecate `try!` macro)
- #62950 (Check rustbook links on all platforms when running locally)
- #63114 (Remove gensym in format_args)
- #63397 (Add tests for some ICEs)
- #63403 (Improve test output)
- #63404 (enable flt2dec tests in Miri)
- #63407 (reduce some test sizes in Miri)
Rollup merge of #63114 - matthewjasper:hygienic-format-args, r=petrochenkov
Remove gensym in format_args
This also fixes some things to allow us to export opaque macros from libcore:
* Don't consider items that are only reachable through opaque macros as public/exported (so they aren't linted as needing docs)
* Mark private items reachable from the root of libcore as unstable - they are now reachable (in principle) in other crates via macros in libcore
bors [Thu, 8 Aug 2019 23:39:01 +0000 (23:39 +0000)]
Auto merge of #63395 - Centril:rollup-kt805cj, r=Centril
Rollup of 6 pull requests
Successful merges:
- #63162 (Miri tests: use xargo to build separate libstd)
- #63289 (Don't recommend `extern crate` syntax)
- #63373 (gitignore: add comment explaining policy)
- #63374 (move of packed fields might or might not occur when they actually are sufficiently aligned)
- #63381 (reduce visibility)
- #63387 (Test interaction between `async { ... }` and `?`, `return`, and `break`)
Rollup merge of #63387 - Centril:async-block-control-flow-tests, r=cramertj
Test interaction between `async { ... }` and `?`, `return`, and `break`
Per the second checkbox in https://github.com/rust-lang/rust/issues/62121#issuecomment-506884048, test that `async { .. }` blocks:
1. do not allow `break` expressions.
2. get targeted by `return` and not the parent function.
3. get targeted by `?` and not the parent function.
Rollup merge of #63374 - RalfJung:pin-packed, r=cramertj
move of packed fields might or might not occur when they actually are sufficiently aligned
See https://github.com/taiki-e/pin-project/pull/34, where it was pointed out that we actually don't move fields of 1-aligned types when dropping a packed struct -- but e.g. in a `packed(2)` struct, we don't do something similar for 2-aligned types. The code for that is [here](https://github.com/rust-lang/rust/blob/db7c773a6be2f050d1d1504763819ea3916f5428/src/librustc_mir/util/alignment.rs#L7).
Rollup merge of #63373 - RalfJung:gitignore, r=alexcrichton
gitignore: add comment explaining policy
Based on https://github.com/rust-lang/rust/pull/63307#issuecomment-518539503, I added a comment what I think should be gitignored and what not. This is just a proposal, obviously. Also see https://github.com/rust-lang/rust/pull/53768 for some more discussion.
The summary is that if there are junk files that you create locally and are fine leaving around (such as `mir_dump`), git has the option for you to add them to `.git/info/exclude`. Others might prefer to keep their working dir clean of those same junk files, so we shouldn't just ignore them for everyone.
I then also cleaned up a few more things, but there were many things that I had no idea where they came from so I didn't touch them.
Rollup merge of #63162 - RalfJung:miri-xargo, r=alexcrichton
Miri tests: use xargo to build separate libstd
This uses `cargo miri setup` to prepare the libstd that is used for testing Miri, instead of adjusting the entire bootstrap process to make sure the libstd that already gets built is fit for Miri.
The issue with our current approach is that with `test-miri = true`, libstd and the test suite get built with `--cfg miri`, which e.g. means hashbrown uses no SIMD, and not all things are tested. Such global side-effects seem like footguns waiting to go off.
On the other hand, the new approach means we install xargo as a side-effect of doing `./x.py test src/tools/miri`, which might be surprising, and we also both have to build xargo and another libstd which costs some extra time. Not sure if the tools builders have enough time budget for that. Maybe there is a way to cache xargo?
We have to first first land https://github.com/rust-lang/miri/pull/870 in Miri and then update this PR to include that change (also to get CI to test Miri before bors), but I wanted to get the review started here.
Cc @oli-obk (for Miri) @alexcrichton (for CI) @Mark-Simulacrum (for bootstrap)
bors [Thu, 8 Aug 2019 17:19:22 +0000 (17:19 +0000)]
Auto merge of #63380 - Centril:rollup-tzfhtnu, r=Centril
Rollup of 8 pull requests
Successful merges:
- #63261 (bump rand in libcore/liballoc test suites)
- #63316 (Update rustfmt to 1.4.4)
- #63332 (Add an overflow check in truncate implementation for Unix.)
- #63342 (Don't use remap-path-prefix in dep-info files.)
- #63366 (doc: Fix typo in float from bytes methods)
- #63370 (Fix ICE #63364)
- #63377 (Improved documentation for compile_error!())
- #63379 (Add test for issue 53096)
Rollup merge of #63342 - ehuss:remap-path-dep-info, r=alexcrichton
Don't use remap-path-prefix in dep-info files.
This changes it so that dep-info files do not use remapped paths.
Having remapped paths causes problems with Cargo because if you remap to a nonexistent path (like `/rustc/$HASH` which rustc distributions do), then Cargo's change tracking thinks the files don't exist and will always rebuild.
I don't actually know if this is a good idea. I think it makes sense, but I do not know what the exact requirements are for reproducible builds. I consider these files separate from the binary artifacts generated, and as a build-system helper, so it seems reasonable to me.
I'm also not sure if this needs a test. I'll definitely add one on the cargo side if this is merged.
Rollup merge of #63261 - RalfJung:rand, r=nikomatsakis
bump rand in libcore/liballoc test suites
This pulls in the fix for https://github.com/rust-random/rand/issues/779, which trips Miri when running these test suites.
`SmallRng` (formerly used by libcore) is no longer built by default, it needs a feature gate. I opted to switch to `StdRng` instead. Or should I enable the feature gate?
projects / rust.git / log