Very Computer

Hi,
another question regarding PHIGS-SI (under Linux). When I open a "default
workstation" via open_phigs, also the monitor-task phigsmon is started to
handle expose events and the like. (I know I can avoid this using open_xphigs,
but I would like to not bother with that for the moment).

The bad thing bad thing is, that (only under Linux ?) phigsmon seems n
endless loop to do this and so steels up to 100% CPU time, running at quite
a high priority, clearly slowing down all other task on the system.
Renicing that task helps a lot without a noticable lag in windows update, but I
really wonder if this behaviour is intended or is due to an error in Linux or
incompatibilities with phigsmon running under X11R6 or what ???? (No other
incompatibilities found yet)

Greetings

Joachim

I don't use PHIGS, so this may not be relevant, but I haven't been
able to get select() to function properly under Linux.  I have a
program that forks the event loop off in a child process, and tries to
use select() to know when to wake up (for inter-process communication
and to handle events).  Since select() doesn't work right, I end up
looping constantly through the event loop, which just burns CPU
cycles.  Your problem may be similar.

-mike

Quote:> Hi,
> another question regarding PHIGS-SI (under Linux). When I open a "default
> workstation" via open_phigs, also the monitor-task phigsmon is started to
> handle expose events and the like. (I know I can avoid this using open_xphigs,
> but I would like to not bother with that for the moment).

> The bad thing bad thing is, that (only under Linux ?) phigsmon seems n
> endless loop to do this and so steels up to 100% CPU time, running at quite
> a high priority, clearly slowing down all other task on the system.
> Renicing that task helps a lot without a noticable lag in windows update, but I
> really wonder if this behaviour is intended or is due to an error in Linux or
> incompatibilities with phigsmon running under X11R6 or what ???? (No other
> incompatibilities found yet)

I'm a bit fuzzy on the PHIGS-SI nowadays, but I remember seeing this
problem once before.  If you look at the source code for phigsmon, you'll
see a number of #ifdefs that control code generation for different
flavors of IPC to be used between the phigsmon process and the application
process.  Try using differnt settings (in the makefile) and recompiling
phigsmon and perhaps the PHIGS lib.  This particular code is very OS-dependent.
It is pretty common for distributors of the PHIGS-SI to not
set up these particular defines correctly.  I can't tell you the right settings
because I do not work with Linux.