Snaipe

I don't think this is necessarily the right fix -- in particular, I think that, semantically speaking, fclosing the returned FILE* handle is the right thing to do from inside...

OK, so the main issue is that `cr_assert_stderr_eq_str` and friends are in fact closing the input, which kind of violates all expectation about how `FILE*` should be used. The right...

Yeah, that happens due to the way that exit status get validated: skipping a test makes the test exit with status 0, while c::should_skip expects to die with SIGSEGV. I'd...

> why not exit with attended value ? If a test want the exit code to be 3, a skip would exit with code 3 It's dicier than that because...

Yeah, I don't think it's a bug (except the fclose crash). cr_assert_stderr_eq_str will consume all of stderr, if you need to assert that the next 3 bytes of stderr are...

This happens because of ASLR. I'm not even sure how to tackle that seamlessly to be honest, because we can't just translate these addresses between the parent and the child...

Do you have a self-contained example that exposes your hanging problems? This aside, letting the user define a max timeout value could be useful, but I'm somewhat unconvinced that this...

Oh, sorry, I thought you were having hanging issues for some of your tests when you upgraded criterion, and wanted to know if you had any reproducer. I definitely interpreted...

Moving this to the v3.0.0 milestone to rename --timeout to --default-timeout, since we can't rename flags without breaking backward-compatibility.

It's technically appropriate -- cr_malloc is for memory allocated in the runner that need to be inherited in the test workers -- but indeed questionable for cases like these. Maybe...