Re: [AD] XIM patch for Allegro 4.1.x

[Elias Pschernig <elias@xxxxxxxxxx>]

On Tue, 2004-10-26 at 20:03 +0200, Elias Pschernig wrote:
> > 
> > Cool!  I will test it out some more later, but I saw a problem in 
> > exkeys: modifier keys should not stuff characters into the readkey() buffer.
> > 
> 
> Hm, you're right. I'll look into it.

Updated the patch to make this work. It should apply against updated
CVS.

> > >I also had to fix a small bug in xdga2.c where the mode fetching called
> > >X11 functions without proper locking.
> > >
> > 
> > You can probably commit this part right now.
> > 
> 
> Will do.
> 

Committed now, along with adding the auto-repeat hack from xwin.c - I
guess DGA needs that as well if X11 does.


> > We need a solution for this.  I think the vsync-emulation could run off 
> > the bgman thread instead of the timer thread.  This will also deal with 
> > the problem of the vsync-emulation not working if the timer subsystem is 
> > not installed.
> 
> It's interesting that this doesn't happen with current CVS. Maybe it is
> connected to now using the MMX for color conversion. I don't know which
> of the examples triggered it.. I was going through all of them to test
> the XIM patch, and one example triggered it. I'll put back vsync and see
> which one it is.
> 

Ok, I couldn't trigger it again - but it can be forced by adding:

install_mouse ();
show_mouse (screen);

to expal.c. It will deadlock immediately, and gdb then shows that just
what I said is happening.

If vsync runs from the background thread, wouldn't it still need access
to the timer counter?

-- 
Elias Pschernig

Attachment: xim2.diff.gz
Description: GNU Zip compressed data