R: [AD] Some news from the QNX port

["Angelo Mottola" <a.mottola@xxxxxxxxxx>]

> Hmm... there are some strange things in the QNX code: the code inits
the
> _bg_man_pthread() driver but it doesn't register any event handler (it

> has its own event handling thread). Moreover there are big
synchronization 
> issues relating to timers.
> Hence my questions: did the QNX port originally use signals ? Did
anyone 
> modify the QNX port after Angelo had brought it to life, in order to
make 
> it use the new pthread framework ?

Yes, the QNX port originally used signals, but for some reasons there
were sync problems also with that method... Anyway the time I started
the port the Allegro pthread API still wasn't coded; it got released
just after I implemented the signals based synchronization method like
in X. In order to solve the sync issues, later I started using pthreads
(not the Allegro API) as well for QNX, and even more later the pthreads
Allegro API, but then I didn't finish as I restarted working on the BeOS
port due to some bug reports and our discussions on priorities for the
4.0 release (we're talking about several months ago!). The Allegro
pthread API also changed during that period, so my work on the QNX side
of it is now really outdated.
So yes, I know QNX Allegro has big synchronization problems... Solving
those will probably mean solving the major stopper for the port.

Angelo Mottola
a.mottola@xxxxxxxxxx
http://www.ecplusplus.com