emixa-d

> I tried to build (NetBSD 9) and got a lot of errors from autogen.sh, including > > ``` > autoreconf: running: aclocal --force -I m4 > autoreconf: running: /usr/pkg/bin/autoconf...

Maybe we could have separate `make-thread-local-fluid` and `make-fiber-local-fluid`? I never use thread local fluids though, so I don't know whether this would be useful.

@attila-lendvai That's not a thread-local fluid (in this case, parameter), but that does sound vaguely like something that could be caused by Fibers useof 'with-dynamic-state/current-dynamic-state'. (Still sounds weird though, will...

@attila-lendvai here is some code testing the situation you are describing ... it doesn't reproduce it, I propose minimizing the test case. ``` (use-modules (fibers) (fibers timers)) (define p (make-parameter...

Hypothesis: shepherd installs a signal handler, which in turn tries to invoke the process monitor. But the signal was sent to a thread that isn't the main thread (created before...

You can test this by (pk (current-thread)) in the main thread, inside the (run-fibers ...) form (should be redundant, but maybe test it just in case), and just before reading...

> the symptom is that even though the entire daemon is wrapped in the dynamic extent of a `parameterize` that sets this fluid, So, I looked at `shepherd.scm` (shepherd 0.10.2),...

Looking at the definition of `handle-SIGCHLD`, the first paragraph appears to be the case. So, seems like a pure Shepherd bug. Also, performing Fibers operations inside an async is super...

(To be clear, I think that being able to do 'put-message' from an async would be great, the issue is that this is _not_ yet actually supported & implemented --...

> @emixa-d is there any facility through which an async signal handler could tell some fiber in the fiber universe to wake up? the handler could package up the signal...