Corentin Schreiber

Since this wasn't receiving more feedback, I have assumed all is well and merged it! If there are still issues to resolve, please feel free to re-open this issue. Thanks...

From what I can piece together: In *Catch2* the equivalent to our `get_current_test()` is called `getResultCapture()`, which in turn calls `getCurrentContext().getResultCapture()`. This current context is a `static` class member (lazily...

The solution adopted in *Catch2*/*doctest*/*GoogleTest* (global test instance) solves the problem of having to propagate the currently running test to child threads. However, it still has issues with asynchronous checks....

It's also worth noting that `REQUIRE(...)` checks, which are meant to immediately stop the test on failure, are inherently difficult to use within a separate thread. It would require a...

Another problem: what about `CAPTURE` and `SECTION`? With the current architecture, these are stored alongside the test state, and would be shared by all threads pointing to that test. This...

> But maybe we don't need the full `shared_ptr` interface; a raw pointer to a stack object and a single atomic ref count would be sufficient (we don't need weak...

Thanks for sharing your reasoning, that is very interesting! > Option A is incompatible with lot of current reporters. That is a good point. It should be possible to adapt...

Hi and thanks for starting this! > whether the zip file inside of zip file is intentional :facepalm: No :( I didn't realize this was happening, thanks for pointing that...

> I was hoping we could simply fetch and upload all the artifacts from the "cmake" workflow, which should include all the platforms. Presumably this can be done in a...

It works! :tada: Thank you