emixa-d

This time I got different failures: `test::check_array_ref_out_of_bounds_5` and `test::check_array_mut_ref_out_of_bounds_32`.

I'd guess so! I guess the first argument of `socket` should be the first argument of `make-default-socket`, feel free to send a patch if you can confirm changing it works...

> printf("Max events fired. Ignoring\n"); I don't think that applications expect fibers to write to stdout. `(current-error-port)` would probably be fine though.

Also, can this ignoring cause some events to be missed?

If we add a fallback for `clock_nanosleep`, shouldn't it be static (maybe not possible with the FFI) or at least namespaces (e.g., _fibers_clock_nanosleep or such)? Otherwise, there's a potential risk...

I don't think unconditionally adding a replacement for `clock_nanosleep` is good -- what if Darwin improves later and adds a `clock_nanosleep`? I think some `AC_CHECK_FUN` would be appropriate.

Looking at my local Guile REPL, we already have 'getaffinity' / 'setaffinity'.

Can't say I I understand this well enough to properly review it, but I've enabled the CI and wrote some comments. Looks like it fails for [email protected] because of (ice-9...

Unless maintainers chime in earlier, I intend to merge this after a month (guile-fibers isn't very active, hence a longer period than usual).

This is for version 1.0.0 btw. Didn't test on master.