Re: [AD] Non C program accessing Allegro on OS X

[Jon Rafkind <workmin@xxxxxxxxxx>]

Peter Hull wrote:

> On Wednesday, October 26, 2005, at 03:24  am, Jon Rafkind wrote:
>
> > While hacking the crap out of the OS X code I discovered that a Bus 
> > error would occur if rest(1) was called while the usleep( 100000 / 
> > refresh_rate ) happened. Im about to goto sleep and dont have to to 
> > investigate it now but does it sound like that might be a race 
> > condition?
>
>
> Rest ends up calling usleep anyway; I don't see why calling usleep in 
> 2 threads can be a race condition. Please let me know what you find!
Actually I dont think theres a problem with rest, it was crashing for 
other reasons.

> Regarding your first problem, it is as Elias says. The complication is 
> that events only get delivered to thread 1* so the event handler must 
> be in thread 1, with your code in thread 2. Other threads are used for 
> timers and audio.
I think Im a little confused as to the sequence of events on OS X.
Is it a)
OS X main
osx_bootstrap_ok --> sets up OSX port or something
call user main
user main
allegro_init
start new thread for OS X handling code

Or is it b)
OS X main
start new thread for OS X handling code
osx_bootrap_ok
call user main
user main
allegro_init

I mean at what point does the OS X code get called? I tried calling the 
OS X setup code by hand but this doesnt work. I think things would work 
if the OS X init code occured somewhere during install_allegro, no?