Martin Nordholts

Thank you for making a follow-up PR! You have real bad luck with CI because you yet again hit a failure not caused by your changes. If you rebase on...

Just a tip in case you didn't know: You can run `./scripts/run-ci-locally.sh` to see if CI will pass. 99% of the time the script is accurate. I would say 100%...

Thanks for the suggestion! Sounds good to me! I think we should wait with implementing until https://github.com/Enselic/cargo-public-api/pull/72 has landed, because otherwise the risk for code conflicts is high. Once the...

`--diff-dirty` sounds good, at least as a preliminary name. For diffing worktree against an arbitrary commit, maybe we could use `--git-diff-checkouts `, i.e. with one commit instead of two? Or...

Making `git stash`/`git stash pop` robust (restoring properly in case of failures) is actually probably quite tricky, so I'm removing the "good first issue" label for now. If we get...

Maybe JSON should have a DynTrait object too? Like in https://github.com/rust-lang/rust/commit/4ea27484c9582389cded2301d1f0dd4f421c8c35

I'm pretty sure we can't solve this in a reasonable way without changes to rustdoc JSON, but I would say that is blocked by https://github.com/Enselic/public-api/issues/88, including the upstream fix. Because...

Right now upstream does not even know of this problem, so I am removing "waiting-on-upstream" for now. I plan on working on this in the near future.

Filed upstream as https://github.com/rust-lang/rust/issues/99118. I will create a PR for upstream in the near future, unless someone else beats me to it.

cargo doc HTML does not present these in isolation, but as part of `impl`s. We show this: ```bash % cd test-apis/example_api-v0.1.0 % cargo public-api --with-blanket-implementations [...] pub type example_api::Struct::Error =...