Check callbacks lifetime

[gwenn]

No description provided.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.57%. Comparing base (8421769) to head (e06566c).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1052   +/-   ##
=======================================
  Coverage   82.57%   82.57%           
=======================================
  Files          53       53           
  Lines       10306    10306           
=======================================
  Hits         8510     8510           
  Misses       1796     1796           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[thomcc]

I actually think this is a bad change. I don't think UnwindSafe works well for FFI code — we only have catch_unwind because it's currently UB to panic over the extern "C" boundary. That's being fixed, and when it gets fixed, we'll remove the catch_unwind.

[gwenn]

I actually think this is a bad change. I don't think UnwindSafe works well for FFI code — we only have catch_unwind because it's currently UB to panic over the extern "C" boundary. That's being fixed, and when it gets fixed, we'll remove the catch_unwind.

Ok, so we should remove all UnwindSafe / RefUnwindSafe constraints (progress_handler already has this contraint) ?
While adding these doc tests, rustc raised a warning that a mut bool is not UnwindSafe (if I remember correctly) which seems useful.

[gwenn]

std library seems to use both (RefUnwindSafe and catch_unwind) here:
https://github.com/rust-lang/rust/blob/master/library/std/src/rt.rs#L107
https://github.com/rust-lang/rust/blob/master/library/std/src/rt.rs#L128

Same here:
https://github.com/Michael-F-Bryan/ffi_helpers/blob/da298c854ad3843f82b82c71900643fbb64dd7dc/src/panic.rs#L43-L45