nanogui: Funny behaviour of nano-X
Subject:
Funny behaviour of nano-X
From:
"Alexander Peuchert" ####@####.####
Date:
2 Jul 1999 06:10:58 -0000
Message-Id: <412567A2.0026CC51.00@nsgdbk01.deutsche-boerse.de>
Hi Gerg, hi Alex,
this night I hacked a little on nano-X including the GGI drivers. I found a
strange behaviour there.
After implementing the basic GGI things the server worked fine. ( using a
select although GGI provides no
FDs for input, so there were no input, but clients could connect and do
some funny drawings...).
Then I changed main.c to poll on the GGI inputs. Not much, just some #if
statements deciding whether polling
should be used or not, then a mousedev.poll() and that functions only calls
a ggiPoll...(). That isn't much, especially
if the mousedev.read() calls the same ggiPoll...() to see if there is an
event pending.
From then on, mouse and keyboard input worked fine( I had to remove mouse
acceleration, though), but many GsRead
errors occured and the demo progam, formerly running smoothly, crashed
after issueing 3-10 nano-X commands.
Also my toolkit could segfault the server calling GrWindowMove,
GrWindowUnMap, GrWindowResize, ... !
Now, where do you think is the problem based: my work, the nano-X windowing
part, network communication, ... ?
- alex