Remove the `ty` field from type system `Const`s
Fixes#125556Fixes#122908
Part of the work on `adt_const_params`/`generic_const_param_types`/`min_generic_const_exprs`/generally making the compiler nicer. cc rust-lang/project-const-generics#44
Please review commit-by-commit otherwise I wasted a lot of time not just squashing this into a giant mess (and also it'll be SO much nicer because theres a lot of fluff changes mixed in with other more careful changes if looking via File Changes
---
Why do this?
- The `ty` field keeps causing ICEs and weird behaviour due to it either being treated as "part of the const" or it being forgotten about leading to ICEs.
- As we move forward with `adt_const_params` and a potential `min_generic_const_exprs` it's going to become more complex to actually lower the correct `Ty<'tcx>`
- It muddles the idea behind how we check `Const` arguments have the correct type. By having the `ty` field it may seem like we ought to be relating it when we relate two types, or that its generally important information about the `Const`.
- Brings the compiler more in line with `a-mir-formality` as that also tracks the type of type system `Const`s via `ConstArgHasType` bounds in the env instead of on the `Const` itself.
- A lot of stuff is a lot nicer when you dont have to pass around the type of a const lol. Everywhere we construct `Const` is now significantly nicer 😅
See #125671's description for some more information about the `ty` field
---
General summary of changes in this PR:
- Add `Ty` to `ConstKind::Value` as otherwise there is no way to implement `ConstArgHasType` to ensure that const arguments are correctly typed for the parameter when we stop creating anon consts for all const args. It's also just incredibly difficult/annoying to thread the correct `Ty` around to a bunch of ctfe functions otherwise.
- Fully implement `ConstArgHasType` in both the old and new solver. Since it now has no reliance on the `ty` field it serves its originally intended purpose of being able to act as a double check that trait vs impls have correctly typed const parameters. It also will now be able to be responsible for checking types of const arguments to parameters under `min_generic_const_exprs`.
- Add `Ty` to `mir::Const::Ty`. I dont have a great understanding of why mir constants are setup like this to be honest. Regardless they need to be able to determine the type of the const and the easiest way to make this happen was to simply store the `Ty` along side the `ty::Const`. Maybe we can do better here in the future but I'd have to spend way more time looking at everywhere we use `mir::Const`.
- rustdoc has its own `Const` which also has a `ty` field. It was relatively easy to remove this.
---
r? `@lcnr` `@compiler-errors`
Add translation support by mdbook-i18n-helpers to bootstrap
This PR add translation support by [mdbook-i18n-helpers](https://github.com/google/mdbook-i18n-helpers) to bootstrap.
This is draft PR because there is the dependency to my forked mdbook-i18n-helpers.
If this PR is acceptable, I'll send a PR to the original mdbook-i18n-helpers and remove draft after changing the dependency to the original.
Closes#124641
Rollup of 9 pull requests
Successful merges:
- #124840 (resolve: mark it undetermined if single import is not has any bindings)
- #125622 (Winnow private method candidates instead of assuming any candidate of the right name will apply)
- #125648 (Remove unused(?) `~/rustsrc` folder from docker script)
- #125672 (Add more ABI test cases to miri (RFC 3391))
- #125800 (Fix `mut` static task queue in SGX target)
- #125871 (Orphanck[old solver]: Consider opaque types to never cover type parameters)
- #125893 (Handle all GVN binops in a single place.)
- #126008 (Port `tests/run-make-fulldeps/issue-19371` to ui-fulldeps)
- #126032 (Update description of the `IsTerminal` example)
r? `@ghost`
`@rustbot` modify labels: rollup
Port `tests/run-make-fulldeps/issue-19371` to ui-fulldeps
This test can run as an ordinary `tests/ui-fulldeps` test, with the help of some additional header variable substitutions to supply a sysroot and linker.
---
Unlike #125973, this test appears to be testing something vaguely useful and breakable, which is why I didn't just delete it.
Winnow private method candidates instead of assuming any candidate of the right name will apply
partially reverts https://github.com/rust-lang/rust/pull/60721
My original motivation was just to avoid the `delay_span_bug` (by attempting to thread the `ErrorGuaranteed` through to here). But then I realized that the error message is wrong. It refers to the `Foo<A>::foo` instead of `Foo<B>::foo`. This is almost invisible, because both functions are the same, but on different lines, so `-Zui-testing` makes it so the test is the same no matter which of these two functions is referenced.
But there's a much more obvious bug: If `Foo<B>` does not have a `foo` method at all, but `Foo<A>` has a private `foo` method, then we'll refer to that one. This has now been fixed, and we report a normal `method not found` error.
The way this is done is by creating a list of all possible private functions (just like we create a list of the public functions that can actually be called), and then winnowing it by analyzing where bounds and `Self` types to see if any of the found methods can actually apply (again, just like with the list of public functions).
I wonder if there is room for doing the same thing with unstable functions instead of running all of method resolution twice.
r? ``@compiler-errors`` for method resolution stuff
Update cargo
9 commits in 7a6fad0984d28c8330974636972aa296b67c4513..34a6a87d8a2330d8c9d578f927489689328a652d
2024-05-31 22:26:03 +0000 to 2024-06-04 15:31:01 +0000
- Silence the warning about forgetting the vendoring (rust-lang/cargo#13886)
- fix(vendor): Ensure sort happens for vendor (rust-lang/cargo#14004)
- fix(add): Avoid escaping double-quotes by using string literals (rust-lang/cargo#14006)
- refactor(source): Split `RecursivePathSource` out of `PathSource` (rust-lang/cargo#13993)
- doc: Add README for resolver-tests (rust-lang/cargo#13977)
- Allows the default git/gitoxide configuration to be obtained from the ENV and config (rust-lang/cargo#13687)
- refactor: Transition direct assertions from cargo-test-support to snapbox (rust-lang/cargo#13980)
- Fix: Skip deserialization of unrelated fields with overlapping name (rust-lang/cargo#14000)
- chore(deps): update alpine docker tag to v3.20 (rust-lang/cargo#13996)
r? ghost
`rustc_parse` top-level cleanups
A bunch of improvements in and around `compiler/rustc_parse/src/lib.rs`. Many of the changes streamline the API in that file from this (12 functions and one macro):
```
name args return type
---- ---- -----------
panictry_buffer! Result<T, Vec<Diag>> T
pub parse_crate_from_file path PResult<Crate>
pub parse_crate_attrs_from_file path PResult<AttrVec>
pub parse_crate_from_source_str name,src PResult<Crate>
pub parse_crate_attrs_from_source_str name,src PResult<AttrVec>
pub new_parser_from_source_str name,src Parser
pub maybe_new_parser_from_source_str name,src Result<Parser, Vec<Diag>>
pub new_parser_from_file path,error_sp Parser
maybe_source_file_to_parser srcfile Result<Parser, Vec<Diag>>
pub parse_stream_from_source_str name,src,override_sp TokenStream
pub source_file_to_stream srcfile,override_sp TokenStream
maybe_file_to_stream srcfile,override_sp Result<TokenStream, Vec<Diag>>
pub stream_to_parser stream,subparser_name Parser
```
to this:
```
name args return type
---- ---- -----------
unwrap_or_emit_fatal Result<T, Vec<Diag>> T
pub new_parser_from_source_str name,src Result<Parser, Vec<Diag>>
pub new_parser_from_file path,error_sp Result<Parser, Vec<Diag>>
new_parser_from_source_file srcfile Result<Parser, Vec<Diag>>
pub source_str_to_stream name,src,override_sp Result<TokenStream, Vec<Diag>>
source_file_to_stream srcfile,override_sp Result<TokenStream, Vec<Diag>>
```
I found the old API quite confusing, with lots of similar-sounding function names and no clear structure. I think the new API is much better.
r? `@spastorino`
Rewrite `suspicious-library`, `resolve-rename` and `incr-prev-body-beyond-eof` `run-make` tests in `rmake.rs` format
Part of #121876 and the associated [Google Summer of Code project](https://blog.rust-lang.org/2024/05/01/gsoc-2024-selected-projects.html).
Some oddly specific ignore flags in `incr-prev-body-beyond-eof`:
```rs
// ignore-none
// ignore-nvptx64-nvidia-cuda
```
it could be interesting to run a try job, but it seems there is no nvidia-cuda in the CI settings (`jobs.yml`).
try-job: armhf-gnu
The `Input::File` and `Input::Text` cases should be very similar.
However, currently the `Input::File` case uses `catch_unwind` because,
until recently (#125815) there was a fallible version of
`new_parser_from_source_str` but only an infallible version of
`new_parser_from_file`. This difference wasn't fundamental, just an
overlooked gap in the API of `rustc_parse`.
Both of those operations are now fallible, so the `Input::File` and
`Input::Text` cases can made more similar, with no need for
`catch_unwind`. This also lets us simplify an `Option<Vec<Diag>>` to
`Vec<Diag>`.
Currently we have an awkward mix of fallible and infallible functions:
```
new_parser_from_source_str
maybe_new_parser_from_source_str
new_parser_from_file
(maybe_new_parser_from_file) // missing
(new_parser_from_source_file) // missing
maybe_new_parser_from_source_file
source_str_to_stream
maybe_source_file_to_stream
```
We could add the two missing functions, but instead this commit removes
of all the infallible ones and renames the fallible ones leaving us with
these which are all fallible:
```
new_parser_from_source_str
new_parser_from_file
new_parser_from_source_file
source_str_to_stream
source_file_to_stream
```
This requires making `unwrap_or_emit_fatal` public so callers of
formerly infallible functions can still work.
This does make some of the call sites slightly more verbose, but I think
it's worth it for the simpler API. Also, there are two `catch_unwind`
calls and one `catch_fatal_errors` call in this diff that become
removable thanks this change. (I will do that in a follow-up PR.)
Align `Term` methods with `GenericArg` methods, add `Term::expect_*`
* `Term::ty` -> `Term::as_type`.
* `Term::ct` -> `Term::as_const`.
* Adds `Term::expect_type` and `Term::expect_const`, and uses them in favor of `.ty().unwrap()`, etc.
I could also shorten these to `as_ty` and then do `GenericArg::as_ty` as well, but I do think the `as_` is important to signal that this is a conversion method, and not a getter, like `Const::ty` is.
r? types
Make `WHERE_CLAUSES_OBJECT_SAFETY` a regular object safety violation
#### The issue
In #50781, we have known about unsound `where` clauses in function arguments:
```rust
trait Impossible {}
trait Foo {
fn impossible(&self)
where
Self: Impossible;
}
impl Foo for &() {
fn impossible(&self)
where
Self: Impossible,
{}
}
// `where` clause satisfied for the object, meaning that the function now *looks* callable.
impl Impossible for dyn Foo {}
fn main() {
let x: &dyn Foo = &&();
x.impossible();
}
```
... which currently segfaults at runtime because we try to call a method in the vtable that doesn't exist. :(
#### What did u change
This PR removes the `WHERE_CLAUSES_OBJECT_SAFETY` lint and instead makes it a regular object safety violation. I choose to make this into a hard error immediately rather than a `deny` because of the time that has passed since this lint was authored, and the single (1) regression (see below).
That means that it's OK to mention `where Self: Trait` where clauses in your trait, but making such a trait into a `dyn Trait` object will report an object safety violation just like `where Self: Sized`, etc.
```rust
trait Impossible {}
trait Foo {
fn impossible(&self)
where
Self: Impossible; // <~ This definition is valid, just not object-safe.
}
impl Foo for &() {
fn impossible(&self)
where
Self: Impossible,
{}
}
fn main() {
let x: &dyn Foo = &&(); // <~ THIS is where we emit an error.
}
```
#### Regressions
From a recent crater run, there's only one crate that relies on this behavior: https://github.com/rust-lang/rust/pull/124305#issuecomment-2122381740. The crate looks unmaintained and there seems to be no dependents.
#### Further
We may later choose to relax this (e.g. when the where clause is implied by the supertraits of the trait or something), but this is not something I propose to do in this FCP.
For example, given:
```
trait Tr {
fn f(&self) where Self: Blanket;
}
impl<T: ?Sized> Blanket for T {}
```
Proving that some placeholder `S` implements `S: Blanket` would be sufficient to prove that the same (blanket) impl applies for both `Concerete: Blanket` and `dyn Trait: Blanket`.
Repeating here that I don't think we need to implement this behavior right now.
----
r? lcnr
The flag propagates cargo configs to `rustc-perf --cargo-config`,
which is particularly useful when the environment is air-gapped,
and you want to use the default set of training crates vendored
in the rustc-src tarball.
The `mir!` macro has multiple parts:
- An optional return type annotation.
- A sequence of zero or more local declarations.
- A mandatory starting anonymous basic block, which is brace-delimited.
- A sequence of zero of more additional named basic blocks.
Some `mir!` invocations use braces with a "block" style, like so:
```
mir! {
let _unit: ();
{
let non_copy = S(42);
let ptr = std::ptr::addr_of_mut!(non_copy);
// Inside `callee`, the first argument and `*ptr` are basically
// aliasing places!
Call(_unit = callee(Move(*ptr), ptr), ReturnTo(after_call), UnwindContinue())
}
after_call = {
Return()
}
}
```
Some invocations use parens with a "block" style, like so:
```
mir!(
let x: [i32; 2];
let one: i32;
{
x = [42, 43];
one = 1;
x = [one, 2];
RET = Move(x);
Return()
}
)
```
And some invocations uses parens with a "tighter" style, like so:
```
mir!({
SetDiscriminant(*b, 0);
Return()
})
```
This last style is generally used for cases where just the mandatory
starting basic block is present. Its braces are placed next to the
parens.
This commit changes all `mir!` invocations to use braces with a "block"
style. Why?
- Consistency is good.
- The contents of the invocation is a block of code, so it's odd to use
parens. They are more normally used for function-like macros.
- Most importantly, the next commit will enable rustfmt for
`tests/mir-opt/`. rustfmt is more aggressive about formatting macros
that use parens than macros that use braces. Without this commit's
changes, rustfmt would break a couple of `mir!` macro invocations that
use braces within `tests/mir-opt` by inserting an extraneous comma.
E.g.:
```
mir!(type RET = (i32, bool);, { // extraneous comma after ';'
RET.0 = 1;
RET.1 = true;
Return()
})
```
Switching those `mir!` invocations to use braces avoids that problem,
resulting in this, which is nicer to read as well as being valid
syntax:
```
mir! {
type RET = (i32, bool);
{
RET.0 = 1;
RET.1 = true;
Return()
}
}
```
Improve compiletest expected/not found formatting
compiletest, oh compiletest, you are truly one of the tools in this repository. You're the omnipresent gatekeeper, ensuring that every new change works, doesn't break the world, and is nice. We thank you for your work, for your tests, for your test runs, for your features that help writing tests, for all the stability and and good you have caused. Without you, Rust wouldn't exist as it does, without you, nothing would work, without you, we would all go insane from having changes break and having to test them all by hand. Thank you, compiletest.
but holy shit i fucking hate your stupid debug output so much i simply cannot take this anymore aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
By changing a few magic lines in this file called "runtest.rs", we can cause compiletest to emit nicer messages. This is widely regarded as a good thing. We stop wasting vertical space, allowing more errors to be displayed at once. Additionally, we add colors, which make it so much more pretty *and* gay, both of which are very good and useful.
There's a bit of fuckery needed to get the colors to work. `colored` checks whether stdout is a terminal. We also print to stdout, so that works well.
But.... for some stupid reason that I absolutely refuse to even attempt to debug, stdout is *not* a terminal when executing tests *in a terminal*.
But stderr is >:).
So this just checks whether stderr is a terminal.
If you have a use case where you dump compiletest stdout into a place where colors are not supported while having stderr be a terminal, then I'm sorry for you, but you are gonna get colors and you're gonna like it. Stop it with the usual environment variable, which `colored` also respects by default.
### before (bad, hurts your brain, makes you want to cry)
## after (good, gay, makes you want to cry)
r? jieyouxu said he wants to review the PR
Rollup of 3 pull requests
Successful merges:
- #125311 (Make repr(packed) vectors work with SIMD intrinsics)
- #125849 (Migrate `run-make/emit-named-files` to `rmake.rs`)
- #125851 (Add some more specific checks to the MIR validator)
r? `@ghost`
`@rustbot` modify labels: rollup
compiletest, oh compiletest, you are truly one of the tools in this
repository. You're the omnipresent gatekeeper, ensuring that every new
change works, doesn't break the world, and is nice. We thank you for
your work, for your tests, for your test runs, for your features that
help writing tests, for all the stability and and good you have caused.
Without you, Rust wouldn't exist as it does, without you, nothing would
work, without you, we would all go insane from having changes break and
having to test them all by hand. Thank you, compiletest.
but holy shit i fucking hate your stupid debug output so much i simply
cannot take this anymore aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
By changing a few magic lines in this file called "runtest.rs", we can
cause compiletest to emit nicer messages. This is widely regarded as a
good thing. We stop wasting vertical space, allowing more errors to be
displayed at once. Additionally, we add colors, which make it so much
more pretty *and* gay, both of which are very good and useful.
There's a bit of fuckery needed to get the colors to work. `colored`
checks whether stdout is a terminal. We also print to stdout, so that
works well.
But.... for some stupid reason that I absolutely refuse to even attempt
to debug, stdout is *not* a terminal when executing tests *in a
terminal*.
But stderr is >:).
So this just checks whether stderr is a terminal.
If you have a use case where you dump compiletest stdout into a place
where colors are not supported while having stderr be a terminal, then
I'm sorry for you, but you are gonna get colors and you're gonna like
it. Stop it with the usual environment variable, which `colored` also
respects by default.
When implementing support for rmake.rs, I copied over the `$TMPDIR`
directory logic from the legacy Makefile setup. In doing so, I also
compiled recipe `rmake.rs` into executables which unfortunately are
placed into `$TMPDIR` as well.
This causes a problem on Windows where:
- The `rmake.exe` executable is placed in `$TMPDIR`.
- We run the `rmake.exe` as a process.
- The process uses `rmake.exe` inside `$TMPDIR`.
- Windows prevents the .exe file from being deleted when the process
is still alive.
- The recipe test code tries to `remove_dir_all($TMPDIR)`, which fails
with access denied because `rmake.exe` is still being used.
We fix this by separating the recipe executable and the sratch
directory:
```
base_dir/
rmake.exe
scratch/
```
We construct a base directory, unique to each run-make test, under
which we place rmake.exe alongside a `scratch/` directory. This
`scratch/` directory is what is passed to rmake.rs tests as `$TMPDIR`,
so now `remove_dir_all($TMPDIR)` has a chance to succeed because
it no longer contains `rmake.exe`.
Oops. This was a fun one to try figure out.
