László Vaskó

That msvcrt issue is a red herring, the `subprocess` module actually tries to import it and falls back to the posix implementation: ```python try: import msvcrt import _winapi _mswindows =...

You shouldn't have it set. It would be nice to figure out how the collision happens. We could for sure rename this module if it can cause problems to others...

It's hard to measure that properly because we don't have performance guaranties for the test runners. We could also measure the consumed CPU time to make the results more reproducible...

I am not sure if we should keep these large files in the repo, even if we use LFS as first checkout can take a very long time.

I used https://nbedos.github.io/termtosvg/ in the past but it is unmaintained. Do we need debug output anyway?

I've been wanting to do the same, however the current packaging relies on poetry2nix which afaik shouldn't be used in nixpkgs packages if at all possible. I have a draft...

I've taken a quick look and it is easy to do this way but the `should_trace` function having a `trace_default_resolver` bool argument makes it a bit smelly to me. One...

The reason we didn't want to go this direction, is tat we wanted to be certain not to use the default value in production. That's why we are using a...

This has bitten me just today: from the changelog of glibc 2.34 (emphasis mine): > * Add _SC_MINSIGSTKSZ and _SC_SIGSTKSZ. When _DYNAMIC_STACK_SIZE_SOURCE > or _GNU_SOURCE are defined, **MINSIGSTKSZ and SIGSTKSZ...

I have created a [tool](https://github.com/vlaci/nix-straight.el) which hooks into straight and generates the repository from nix package derivations. I don't know if it is viable as a general purpose tool outside...