| Re: [AD] demo of new timer API (unix only) |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
> >I'm back at any rate. I think Elias and Angelo said they'd be away until
> >the end of the month though...
>
> Greece, perhaps?
Sounds sensible.
> If it's the pthreads package that I looked at in the past, it doesn't
> have preemptive threads, i.e. a thread only gets the CPU when another
> thread calls one of the pthread_* functions. That's not very helpful
> and so the emulation either won't work or will be that much harder.
From what I could gather, there is a pthreads version for DJGPP that does
have preemptive threads. I wasn't able to find an easy download link that
worked to check it out though (then again, I didn't use the zip picker...)
> Nah, it's not that simple :-) It will only get harder with the input
APIs.
Hmm... good point.
> Cheers. I'll put up a newer version tonight as there's a thread-safety
> bug in the previous version I uploaded. Silly Allegro calls
> install_int/remove_int from inside interrupt callbacks...
I didn't experience any problems running your last version here. Everything
seems to be working fine here. I'm guessing that the old timer stuff
should eventually be moved to a compat.c compatibility file? Or may placed
in an src/compat subdirectory once there are more files for which backward
compatibility wrappers are needed?
Oh, a question about threading: what was the concensus on exposing a
multi-threading API from Allegro?
Evert