Eliminate use of #[cfg_attr(not(doc), repr(...))]
#116743
Conversation
|
r? @m-ou-se (rustbot has picked a reviewer for you, use r? to override) |
rustbot
added
S-waiting-on-review
|
Portable SIMD is developed in its own repository. If possible, consider making this change to rust-lang/portable-simd instead. |
dtolnay
force-pushed
the
notdoctransparent
branch
from
27411f9 to
56f472d
Compare
|
|
There was a problem hiding this comment.
This can't land as is, I'm afraid. My PR only affects the transparent representation. I've totally overlooked non-transparent reprs but in my defense nobody has mentioned this either over at the issue and the PR.
It does make sense though to extend this behavior to other reprs 🤦. I will open a PR as soon as possible.
CC @RalfJung
| @@ -204,7 +204,7 @@ mod c_long_definition { | |||
| // be UB. | |||
| #[doc = include_str!("c_void.md")] | |||
| #[lang = "c_void"] | |||
| #[cfg_attr(not(doc), repr(u8))] // work around https://github.com/rust-lang/rust/issues/90435 | |||
| #[repr(u8)] | |||
There was a problem hiding this comment.
Unfortunately, this will still show up in the docs, my PR only affects repr(transparent). Further, it doesn't take into account #[doc(hidden)] (in this case on variants). Arguably, rustdoc should hide #[repr(u8)] and respect doc(hidden). I can send a PR right away.
There was a problem hiding this comment.
Shouldn't we keep non-transparent reprs for enums though? It's useful, especially repr(c) ones for FFI.
There was a problem hiding this comment.
For sure, but in this case both variants are doc(hidden). I plan on hiding those reprs if all variants are hidden.
| @@ -297,7 +297,7 @@ impl<'f> fmt::Debug for VaListImpl<'f> { | |||
| not(target_os = "uefi"), | |||
| not(windows), | |||
| ))] | |||
| #[cfg_attr(not(doc), repr(C))] // work around https://github.com/rust-lang/rust/issues/66401 | |||
| #[repr(C)] | |||
There was a problem hiding this comment.
Similarly, repr(C) is still going to show up in the docs :| Let me update rustdoc to handle that as well.
dtolnay
added
S-blocked
…rk-Simulacrum Request that rust-analyzer changes are sent upstream first if possible This automates `@lnicola's` comment rust-lang#118253 (comment). Rustbot will write a comment similar to rust-lang#116743 (comment).
Rollup merge of rust-lang#118255 - dtolnay:mentionsrustanalyzer, r=Mark-Simulacrum Request that rust-analyzer changes are sent upstream first if possible This automates `@lnicola's` comment rust-lang#118253 (comment). Rustbot will write a comment similar to rust-lang#116743 (comment).
|
I'll rebase and reopen if #116882 makes progress. |
This PR reverts #107680 and some pre-existing occurrences of the same pattern.
The issue that prompted the workaround, #90435, has since been fixed by #115439.
Tested by running
x.py doc --stage 1 library/stdand confirming that std/path/struct.PathBuf.html contains no visible#[repr(transparent)]. Note thatx.py doc library/stdruns using stage0 rustdoc (maybe depending on config.toml?) and will continue to render a#[repr(transparent)], but that's okay and shouldn't make it into official documentation of the 1.75.0 release.Blocked on #116882