Results 27 comments of tottoto

I think this implementation would make the lock drop before panic which occurs when each test fails.

I think `drop` should be called before `resume_unwind`.

Do you have the reproduction which show that they behaves the same?

Thanks, I have known that behavior. In my test, without manually dropping the guard before calling resume_unwind, when one of the tests (which are called in this helper function) fails...

Yes, and more precisely, manual dropping is only necessary when the test fails, so it seems better to move it just before resume_unwind.

Thanks for another solution. Using `parking_lot` sounds good as well. Although in my opinion the mechanism which is implemented does not sound too complicated to exclude by using another crate...

Addressed at #1707. Thank you!

Resolved in #1532. Thank you!

Seems to have been resolved in #3314.

`socket2`'s ESP-IDF support is [backported to 0.4](https://github.com/rust-lang/socket2/pull/455) and [released at 0.4.10](https://github.com/rust-lang/socket2/blob/v0.4.10/CHANGELOG.md#0410).