regression: type annotations required cannot resolve std::borrow::Cow<'_, str>: std::borrow::Borrow<_>

[Mark-Simulacrum]

root: zmtp - 2 detected crates which regressed due to this; cc @jjpe

• rustyq-0.6.0: start v. end; cc @jjpe
• zmtp-0.6.0: start v. end; cc @jjpe

[eddyb]

I can't find a library change that would cause this (and rustdoc confirms that Borrow has the same set of impls on stable and beta), marking as T-compiler.

[pnkfelix]

This regression was injected by commit 1962ade aka PR #60877, which rolled up the following PR's:

• string: implement From<&String> for String #59825 (string: implement From<&String> for String)
• Fix convert module's documentation links #59923 (Fix convert module's documentation links)
• Include expression to wait for to the span of Await #60691 (Include expression to wait for to the span of Await)
• Update rustc book CLI docs. #60769 (Update rustc book CLI docs.)
• README.md: Mention MSVC 2017+, not 2013(!) #60816 (README.md: Mention MSVC 2017+, not 2013(!))
• Move box from the stable keyword to unstable keywords list #60851 (Move box from the stable keyword to unstable keywords list)

[pnkfelix]

(could this regression have been caused by PR #59825 ?)

[pnkfelix]

Apparently so; this appears to be a duplicate of #60958?

Issue #60958 seems has already been discussed amongst the libs team with the conclusion that this is acceptable fallout from PR #59825?

[pnkfelix]

I would file a followup issue against the zmtp crate, but it is hosted on bitbucket.org, and apparently my account credentials there no longer work... its link on crates.io to bitbucket.org is actually dead.

[pnkfelix]

Closing as duplicate of #60958.

[jjpe]

So this will require downstream libs (among which zmtp) to fix code that used to work on stable but no longer will once this rustc beta stabilizes, do I understand that correctly?

[pnkfelix]

So this will require downstream libs (among which zmtp) to fix code that used to work on stable but no longer will once this rustc beta stabilizes, do I understand that correctly?

@jjpe that interpretation is consistent with my understanding. See further discussion on #60958, for example this comment

[pnkfelix]

In particular, changes that cause crates to need additional type annotations are allowed under the stability guarantees as described in this blog post