Missing docs for non-first impl

[Mark-Simulacrum]

e.g., https://doc.rust-lang.org/nightly/std/panic/struct.Location.html is missing the file/line/column method docs which are present in beta https://doc.rust-lang.org/beta/std/panic/struct.Location.html#impl-1

cc @rust-lang/rustdoc

[JohnTitor]

Removing I-prioritize as this is T-rustdoc issue.

[Manishearth]

@JohnTitor Why?

[Manishearth]

A quick test indicates that copying the source of std::panic::Location out into a no_std crate doesn't fix the problem

[Manishearth]

Oh wait I don't have the latest nightly

[JohnTitor]

@Manishearth As mentioned in #72340 (comment), I'm not sure this should be prioritized under t-compiler's wg-prioritization. If this label is useful for the rustdoc team as well, we could tweak triagebot not to notify on Zulip.

[Manishearth]

@JohnTitor Yeah it seems like it would be useful for us

[Manishearth]

Hmm so I am unable to reproduce this bug on a self contained no-core file that contains the exact code of Location aside for changes in the body

[Manishearth]

Regression range 7e11379..5c9e5df

5c9e5df Auto merge of #74408 - Manishearth:rollup-9gxn4od, r=Manishearth
6ee1b62 Auto merge of #72481 - marmeladema:duration-consts-2, r=oli-obk
125c58c Auto merge of #74202 - oli-obk:mir_const, r=RalfJung
4cd0ee9 Auto merge of #74388 - Manishearth:rollup-i7iueu8, r=Manishearth
e2e29de Auto merge of #74375 - Manishearth:rollup-10vbpdh, r=Manishearth

[Manishearth]

Works in 6ee1b62

Regressing PR is #74408

[Manishearth]

Rustdoc PRs in that rollup

fc09817 Rollup merge of #74370 - Manishearth:re-spotlight, r=GuillaumeGomez
61fccf0 Rollup merge of #74359 - lzutao:rustdoc-tostring, r=GuillaumeGomez
196243e Rollup merge of #74325 - GuillaumeGomez:focus-source-file-sidebar, r=kinnison
61a3f67 Rollup merge of #74148 - GuillaumeGomez:doc-alias-check, r=Manishearth
b700835 Rollup merge of #73807 - euclio:rustdoc-highlighting, r=ollie27,GuillaumeGomez
a8bb245 Rollup merge of #73771 - alexcrichton:ignore-unstable, r=estebank,GuillaumeGomez
c23f045 Rollup merge of #73566 - jyn514:name-resolve-first, r=eddyb

[Manishearth]

Confirmed, 9ffb7801363cff45dcb56fa2146a48aa0ae433db (try build from #73771) is the culprit

[Manishearth]

What's happening is that that code erroneously checks the stability of the impl, which apparently is "unstable if it contains unstable items". The "foreign item" check applies to libstd reexporting libcore. It shouldn't.

The minimal reproduction is two crates, one using stability attrs such that an impl has at least one unstable item, and the second crate reexporting the type.

[Mark-Simulacrum]

I'm going to take a stab at fixing this.