Change --extern-html-root-url to use crate source path as key

[Nemo157]

This is the most correct solution I see to #76296, this allows setting the html-root-url for all crates in the dependency tree, no matter how many duplicates of a specific crate name exist.

I have prototyped the associated docs.rs changes in my script that builds in the same fashion as it: Nemo157/dotfiles@d452627. Before this is merged we would want to do the same for docs.rs itself (I'll make a PR for this shortly), it's forwards-compatible to start emitting both sets of flags and remove the current ones later, extra unused flags are just ignored.

fixes #76296

[rust-highfive]

r? @jyn514

(rust_highfive has picked a reviewer for you, use r? to override)

[jyn514]

This seems to expose build details a little bit - does cargo even provide a way to get the .so file that corresponds to a crate? I wouldn't want to make this breaking change without knowing there's a way to actually use it, if you could put together a demo of a cargo-based project that would make me feel a lot better.

cc @eddyb - this changes rustdoc to look up info about a crate by the source path (extern-html-root-url/auxiliary/libextern_html_root_url.so) instead of the crate name (extern_html_root_url) so that rustdoc can distinguish between versions of a crate with the same name.

[Nemo157]

This seems to expose build details a little bit - does cargo even provide a way to get the .so file that corresponds to a crate? I wouldn't want to make this breaking change without knowing there's a way to actually use it, if you could put together a demo of a cargo-based project that would make me feel a lot better.

See the link to my cargo doc-like-docs.rs changes, using cargo check --message-format=json it emits compiler-artifact messages with the files that were produced for each crate.

[Nemo157]

It needs significant cleanup, but rust-lang/docs.rs@master...Nemo157:issue-76296 mostly implements this on the docs.rs side.

example command from docs.rs build log

[ehuss]

Hm, using extern paths is an interesting approach! It does neatly solve the problem of having some unique identifier to correlate each extern crate. Another approach I had thought about was to encode the -C extra-filename in the crate metadata, and then use that hash as an identifier, but that didn't seem very good.

It would be nice to have some strategy to also fix #56169. I'm not sure how this could fit in with that, or if that will need to be solved with a completely different approach.

I would be concerned about comparing paths directly. Using absolute paths tends to be tricky and runs the risk of not working in various situations. same-file would be an option to reliably compare paths, but I'm uncertain if that would have a significant performance impact.

Regarding the concern about long command-line paths, I'm thinking that the best option is to use a response file. I'm a little uncomfortable doing that in Cargo (it tends to be a pain to work with), but I'm not sure if there are many other options.

Even though I am the author of the crate_extern_paths query, I am uncertain if it is appropriate to use it in this fashion.

The documentation should probably be updated.

[jyn514]

r? @ehuss

[Nemo157]

One thought I just had to reduce the command size back a bit; we probably don't need the absolute path, given the metadata suffix just using the filename would probably be enough, so that would only add an extra ~26 characters per crate.

[bors]

☔ The latest upstream changes (presumably #79104) 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

[crlf0710]

Triage: There's merge conflicts now.

[JohnCSimon]

Ping from triage
@Nemo157 - can you please resolve the merge conflict, and alert the reviewer? Thank you!

[Dylan-DPC-zz]

@Nemo157 any updates on this?

[Dylan-DPC-zz]

Closing this as inactive