| Re: [AD] X11 unresponsiveness |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
On Tuesday 13 July 2004 01:36, Elias Pschernig wrote:
> Ok, the attached patch should completely resolve the lag problem for the
> threaded version, and improve it a lot with the signals version. With
> the threaded version, it is like we said - the thread blocks in XLOCK,
> then input is read. Maximum responsiveness is sure.
Yes, this works wonderfully! The demo game actually feels more responsive
than in an unpatched version of Allegro too with this patch applied.
> With the signals version, I'm timing how long it was since the last
> event was process. If it is longer than 50ms, all function calling XLOCK
> start blocking, until the event signal is called again. (We can't just
> process events, or block inside _xwin_event_handler, since it is a
> signal handler and has to return soon.) Of course, this is not optimal,
> since how all that is ensured is, that at least 5 events are processed
> every 50ms. But it as a lot better than before.
With your patches applied, the sigalrm version of the library crashes for
me. I think I'm missing an additional patch.
Evert