This commit removes the custom index implementation of `NodeIndex`,
which probably predates `newtype_index!`.
As well as eliminating code, it improves the debugging experience,
because the custom implementation had the property of being incremented
by 1 (so it could use `NonZeroU32`), which was incredibly confusing if
you didn't expect it.
For some reason, I also had to remove an `unsafe` block marker from
`from_u32_unchecked()` that the compiler said was now unnecessary.
Rollup of 3 pull requests
Successful merges:
- #63872 (Document platform-specific behavior of the iterator returned by std::fs::read_dir)
- #64250 (save-analysis: Nest typeck tables when processing functions/methods)
- #64472 (Don't mark expression with attributes as not needing parentheses)
Failed merges:
r? @ghost
Don't mark expression with attributes as not needing parentheses
This is not perfectly correct as `#[attr] (5)` will still not lint, but it does seem good enough, in particular as the parentheses in that case are not unambiguously incorrect; I might personally prefer to see them for clarity.
Fixes https://github.com/rust-lang/rust/issues/43279.
save-analysis: Nest typeck tables when processing functions/methods
Fixes an issue where we did not nest tables correctly when resolving
associated types in formal argument/return type positions.
This was the minimized reproduction case that I tested the fix on:
```rust
pub trait Trait {
type Assoc;
}
pub struct A;
pub fn func() {
fn _inner1<U: Trait>(_: U::Assoc) {}
fn _inner2<U: Trait>() -> U::Assoc { unimplemented!() }
impl A {
fn _inner1<U: Trait>(self, _: U::Assoc) {}
fn _inner2<U: Trait>(self) -> U::Assoc { unimplemented!() }
}
}
```
using `debug_assertions`-enabled rustc and by additionally passing `-Zsave-analysis`.
Unfortunately the original assertion fired is a *debug* one and from what I can tell we don't run the tests with these on, so I'm not adding a test here. If I missed it and there is a way to run tests with these on, I'd love to add a test case for this.
Closes#63663Closes#50328Closes#43982
Cleanup handling of hygiene for built-in macros
This makes most identifiers generated by built-in macros use def-site hygiene, not only the ones that previously used gensyms.
* `ExtCtxt::ident_of` now takes a `Span` and is preferred to `Ident::{from_str, from_str_and_span}`
* Remove `Span::with_legacy_ctxt`
* `assert` now uses call-site hygiene because it needs to resolve `panic` unhygienically.
* `concat_idents` now uses call-site hygiene because it wouldn't be very useful with def-site hygiene.
* everything else is moved to def-site hygiene
r? @petrochenkov
