feat: Object safe trait nightly test

[devanbenz]

• Adds test_crate and output for trait_no_longer_object_safe lint
• This is a +nightly rustdoc feature currently so will remain open as a draft PR until the feature is moved in to stable

[obi1kenobi]

Reminder to self:

• Rust 1.77 is the first one to ship trait dyn-safety in rustdoc JSON.
• The test cases here were checked and valid on Rust 1.77 beta and 1.78 nightly circa 2024-02-11. They are extremely unlikely to drift.
• Merge this PR when Rust 1.77 is our lower bound.

[obi1kenobi]

@devanbenz if you don't mind, could you remove the lint file and macro invocation change, and add the trailing newlines to all the test files here?

[devanbenz]

Yep absolutely!

…

On Mon, Feb 12, 2024 at 6:59 PM Predrag Gruevski ***@***.***> wrote: @devanbenz <https://github.com/devanbenz> if you don't mind, could you remove the lint file and macro invocation change, and add the trailing newlines to all the test files here? — Reply to this email directly, view it on GitHub <#661 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFLL2B5A5VI77E2IRFPVDK3YTK3GJAVCNFSM6AAAAABDFTERGCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZZHE3DSOBZGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[obi1kenobi]

Looks good, thanks!

[devanbenz]

https://releases.rs/ - when will 1.77 be lower bound? I know as of right now its stable.

[obi1kenobi]

Right now I'm thinking probably in another 3-4 Rust releases or so (~4-6 months), unless there's a huge burst of breaking changes in rustdoc JSON formats in consecutive Rust releases that causes more maintenance burden.

Any particular reason you're asking?

I'm under the impression that only the test case is left outstanding, all the rest of the work is merged and the Rust community already benefits from the new lint if they use Rust 1.77+. Is that also how you see it? Just wanted to make sure I'm not making some incorrect starting assumption here.

[devanbenz]

Right now I'm thinking probably in another 3-4 Rust releases or so (~4-6 months), unless there's a huge burst of breaking changes in rustdoc JSON formats in consecutive Rust releases that causes more maintenance burden.

Any particular reason you're asking?

I'm under the impression that only the test case is left outstanding, all the rest of the work is merged and the Rust community already benefits from the new lint if they use Rust 1.77+. Is that also how you see it? Just wanted to make sure I'm not making some incorrect starting assumption here.

Nope, no specific reason! I've been just doing some spring cleaning on my PRs and issues in GitHub 😆

[obi1kenobi]

Thanks for keeping an eye on this and checking in!

[obi1kenobi]

I just released v0.33.0 of cargo-semver-checks, and that's the last release that will support Rust older than 1.77. After the release, I also merged a PR to drop Rust versions prior to 1.77 from the test matrix.

That means this PR is finally ready to merge!

Thanks for your patience, and for helping build this functionality!