Warn on broken intra-doc links added to cross-crate re-exports

[GuillaumeGomez]

This emits broken_intra_doc_links for docs applied to pub use statements that point to external items and are inlined.
Does not address #77200 - any existing broken links from the original crate will not show warnings.

r? @jyn514

[jyn514]

@GuillaumeGomez can you add an example with a broken link on the original docs, before it was re-exported? Like this:

// crate1/// [not_found]pubfnf(){}// crate2pubuse crate1::f;

cc @rust-lang/rustdoc, this is a change in behavior and undoes #56922
cc @euclio

[GuillaumeGomez]

Sure! Just to be sure: we don't want a warning in this case, right?

[jyn514]

I'm not sure about that. Consider the examples from #77200:

• cfg(doc) on an import, so that the link only fails when re-exported
• crate::x meant something different depending on the crate, so it only failed when re-exported (fixed in Resolve crate in intra-doc links properly across crates #77253)

As a library author I would want to know about those I think, instead of having them silently fail. At least then you could work around it somehow.

[jyn514]

That said, I'm fine with only reporting lints on new documentation for now so we can get the fix in, this would unblock #77254.

[Manishearth]

As a library author I would want to know about those I think, instead of having them silently fail. At least then you could work around it somehow.

I'm wary of reporting bugs that take updating a dependency to fix. As long as it gets reported in the dependency (which will show up in CI logs!) it should be fine IMO

[jyn514]

I'm wary of reporting bugs that take updating a dependency to fix. As long as it gets reported in the dependency (which will show up in CI logs!) it should be fine IMO

Well that's the issue, some of these won't show up in the dependency. So there's no way to see them other than warning on downstream crates, and we're currently silencing those warnings.

[GuillaumeGomez]

Updated!

[jyn514]

@GuillaumeGomez just double checking - this only affects intra-doc links for now, right? Everywhere else ignores use statements? If so r=me and we can debate whether this should lint the original warning in re-exports later (currently it does not): #77276 (comment)

[GuillaumeGomez]

Only intra-doc links as far as I can tell. But all other lints are now able to run checks on them as well, as long as they want to.

@bors: r=jyn514

[bors]

📌 Commit 1f80f981ae4b56d3f91d81432c41d0b25eff2eae has been approved by jyn514

[ollie27]

This now displays re-exports for items even when they have been inlined. Here's std for example:

That can't be intentional, right?

[jyn514]

@bors r-

No, that wasn't intentional. @GuillaumeGomez can you fix that?

[GuillaumeGomez]

So I just need to check for doc(inline) I assume? I'll do that tomorrow then.

[jyn514]

So I just need to check for doc(inline)

Well, I think the fix needs to happen in render, not in clean. Some parts of rustdoc (intra-doc links) are interested in the use and some (render) are not. So each needs to decide on their own whether to ignore it or not.

[Manishearth]

Well that's the issue, some of these won't show up in the dependency. So there's no way to see them other than warning on downstream crates, and we're currently silencing those warnings.

Wait, why not? Can you give me an example?

[jyn514]

Wait, why not? Can you give me an example?

#77276 (comment): #77193, #75855. The first one is just a rustdoc bug, but it's possible there are more like it. The second one is a legitimate difference between documenting normally and documenting as a dependency.

[GuillaumeGomez]

Just waiting on you then @ollie27. :)

[bors]

☔ The latest upstream changes (presumably #77519) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[jyn514]

#77254 is fixed so people can now fix the warnings they get from broken links :)

[GuillaumeGomez]

Since I removed the filter on compiler added imports, we can move on.

@bors: r=jyn514,ollie27

[bors]

📌 Commit 7e218bb has been approved by jyn514,ollie27

[bors]

⌛ Testing commit 7e218bb with merge 38d911d...

[bors]

☀️ Test successful - checks-actions, checks-azure
Approved by: jyn514,ollie27
Pushing 38d911d to master...