Sebastian Pfitzner
Sebastian Pfitzner
My two cents are that we should treat this as a bug/unexepected behaviour in the flisp parser.
I'd suggest [`flock`ing](https://man7.org/linux/man-pages/man2/flock.2.html) the lockfile instead of just checking for its presence. That neatly addresses the concerns, in that 1. multiple instances of the dyad compiler can actually check whether...
Hm, I can interrupt that ok with inline evaluation, but you have to hit Ctrl-C a few times. It's possible that this hits the "force interrupt" heuristic I added a...
Are you on Windows?
Ah, that explains it then ;) The forced interrupt isn't supported on Windows.
That means that the normal mechanism for scheduling an `InterruptException` on the worker process doesn't work in this case (probably because the loop never `yield`s), but the signal based approach...
The notebook issue is different from this, I think.
I've also gotten very frequently annoyed at this, but properly fixing this would require reworking our code execution mechanism quite a bit -- I don't understand how signals are supposed...
Yeah, probably. Also there's the standard caveat that none of this will work on Windows anyways (which IIRC was the original motivation for implementing the soft interrupt concept).
I can't repro this, although I can imagine this going wrong in some way. What's the actual failure mode here?