…eady tainted

…rs on

This commit refactors the `#[do_not_recommend]` support in the old
parser to also apply to projection errors and not only to selection
errors. This allows the attribute to be used more widely.

Signed-off-by: Andrew Wock <[email protected]>

`TyCtxt` impls `PpAnn` in `compiler/rustc_middle/src/hir/map/mod.rs`. We
can call that impl, which then calls the one on `intravisit::Map`,
instead of calling the one on `intravisit::Map` directly, avoiding a
cast and extra references.

Exhaustiveness and usefulness checking are now in
`rustc_pattern_analysis`.

A lot of errors don't need to be visible outside the crate, and some
other things as well.

Silence follow-up errors directly based on error types and regions

During type_of, we used to just return an error type if there were any errors encountered. This is problematic, because it means a struct declared as `struct Foo<'static>` will end up not finding any inherent or trait impls because those impl blocks' `Self` type will be `{type error}` instead of `Foo<'re_error>`. Now it's the latter, silencing nonsensical follow-up errors about `Foo` not having any methods.

Unfortunately that now allows for new follow-up errors, because borrowck treats `'re_error` as `'static`, causing nonsensical errors about non-error lifetimes not outliving `'static`. So what I also did was to just strip all outlives bounds that borrowck found, thus never letting it check them. There are probably more nuanced ways to do this, but I worried there would be other nonsensical errors if some outlives bounds were missing. Also from the test changes, it looked like an improvement everywhere.

…diganostic_namespace, r=compiler-errors

Refactor `#[diagnostic::do_not_recommend]` support

This commit refactors the `#[do_not_recommend]` support in the old parser to also apply to projection errors and not only to selection errors. This allows the attribute to be used more widely.

Part of rust-lang#51992

r? `@compiler-errors`

<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.

This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using

    r​? <reviewer name>
-->

…-suggestion, r=compiler-errors

Improve renaming suggestion for names with leading underscores

Fixes rust-lang#125650

Before:
```
error[E0425]: cannot find value `p` in this scope
 --> test.rs:2:13
  |
2 |     let _ = p;
  |             ^
  |
help: a local variable with a similar name exists, consider renaming `_p` into `p`
  |
1 | fn a(p: i32) {
  |      ~
```

After:
```
error[E0425]: cannot find value `p` in this scope
 --> test.rs:2:13
  |
1 | fn a(_p: i32) {
  |      -- `_p` defined here
2 |     let _ = p;
  |             ^
  |
help: the leading underscore in `_p` marks it as unused, consider renaming it to `p`
  |
1 | fn a(p: i32) {
  |      ~
```

This change doesn't exactly conform to what was proposed in the issue:

1. I've kept the suggested code instead of solely replacing it with the label
2. I've removed the "...similar name exists..." message instead of relocating to the usage span
3. You could argue that it still isn't completely clear that the change is referring to the definition (not the usage), but I'm not sure how to do this without playing down the fact that the error was caused by the usage of an undefined name.

…ee1-dead

Fix ICE caused by ignoring EffectVars in type inference

Fixes rust-lang#119830
​r? ```@matthiaskrgr```

…s, r=lqd

Streamline `nested` calls.

`TyCtxt` impls `PpAnn` in `compiler/rustc_middle/src/hir/map/mod.rs`. We can call that impl, which then calls the one on `intravisit::Map`, instead of calling the one on `intravisit::Map` directly, avoiding a cast and extra references.

r? `@lqd`

…ps, r=compiler-errors

Reduce `pub` exposure in `rustc_mir_build`

r? compiler

Split smir `Const` into `TyConst` and `MirConst`

Part of rust-lang#125958

Building a `smir::Const` currently requires accessing the `Ty<'tcx>` of a `ty::Const`. This will stop being possible in the future. Replicate the split in rustc of having a representation of type level constants and mir constants with the latter being able to store the former. Ideally we wouldnt have `MirConst::Ty` but 🤷‍♀️

r? `@oli-obk`

Store the types of `ty::Expr` arguments in the `ty::Expr`

Part of rust-lang#125958

In attempting to remove the `ty` field on `Const` it will become necessary to store the `Ty<'tcx>` inside of `Expr<'tcx>`. In order to do this without blowing up the size of `ConstKind`, we start storing the type/const args as `GenericArgs`

r? `@oli-obk`