rustdoc: search results show promising result, but referent page doesn't contain the named method.

[nathan-at-least]

Summary: I search API docs, see a promising result, follow that result to a page, that page does not have the method nor description text that was in the search result. Verified in local docs and officially hosted docs.rs docs.

Details:

I originally filed this issue in the tokio repo thinking it was an API doc bug on their part. Someone suggested it's a rust-doc bug, so here I am. The original issue has a fair amount of detail: tokio-rs/tokio#1738

Reproduction:

I haven't literally tested this yet, but I believe it's on the right track:

$ cargo new docbugdemo
$ cd ./docbugdemo
$ echo 'tokio = "0.2.0-alpha.6" >> Cargo.toml
$ cargo doc --open

-then search for next or _::next, and follow the top hit. On the resulting page use browser find-in-page-search for "next". Observe 0 hits. View the source of that module, use browser find-in-page for "next". Observe 0 hits.

Investigation:

I'm missing a lot of context here with both rust-doc and tokio API, but I wonder if some of these could be contributors to the bug:

• tokio re-exports a lot.
• Maybe tokio defines multiple different _ traits or types in different modules?
• Maybe the name _ happens to confuse rust-doc.

Related Issues:
I searched this repo with label:T-rustdoc search and scanned for duplicates/related items. I believe this bug is unique.

Two relevant results:

• rustdoc: Pain points of reexports #24305 - umbrella issue about re-exports which may encompass this bug. I know tokio has many re-exports but I'm not certain yet if that's an issue here.
• Rustdoc search result links to 404 #63435 - maybe this bug has a similar cause but slightly different conditions?

[jonas-schievink]

This might be a duplicate of #66046. What's the method in the search results you clicked on?

[nathan-at-least]

In the linked original issue, tokio-rs/tokio#1738, there are screen shots that show the highlighted result I followed, which is to tokio::prelude::_::next.

To me it doesn't sound like #66046 is a duplicate, because when I view the source link of the referent page, there's literally no "next" anywhere in the source, so this doesn't seem like a case of a hidden / folded doc item.

[nathan-at-least]

Hey, I just realized there are multiple extension traits imported as _, starting with this line: https://docs.rs/tokio/0.2.0-alpha.6/src/tokio/prelude.rs.html#14

Could it be that this confuses the search indexing?

[jyn514]

Duplicate of #61592, fixed by #80267.