Hi Johannes,
So I do have the nvidia headers installed in /usr/include/GL
But I have not changed the driver version, and was able to compile older CVS versions of Gem via these headers, so I don't think buggy nvidia headers are the issue. So to test if if having troubles because of something changed in Gem, or something changed on my machine I tried to compile gem 0.90.0.
I can get through a few of the Manips, but it bails early at:
c++ -c -g -g -O2 -D__NEW__ -fPIC -mmmx -I../ -I/usr/include/avifile-0.7 -I/usr/include/FTGL/ -I/usr/include/freetype2/ -fno-builtin -O3 -funroll-loops -ffast-math rotate.cpp -o rotate.o ortho.cpp:122: error: parse error at end of input make[1]: *** [ortho.o] Error 1 make[1]: *** Waiting for unfinished jobs.... make[1]: Leaving directory `/home/bbogart/pd-stuff/gem-0.90.0-nocamera/src/Manips' make: *** [subdirs] Error 255
It first bailed on camera, so thinking it may be a fluke I removed it from the build and then ortho gave me the same error.
Maybe this will give a hint as to whats up.
Also I removed the pd-devel stuff (make uninstall) and that made no difference to the CVS build.
Should I move my nvidia GL headers out of the way and reinstall them with the mesa headers?
I have no idea whats going on.
Thanks for helping.
.b.
IOhannes m zmoelnig wrote:
B. Bogart wrote:
Hi Johannes,
IOhannes m zmoelnig wrote:
ok, thanks for the file (sent off-list for those who don't know)
which GL-headers do you have installed? nvidia's? i have heard rumours that they are severly broken; i am using the headers provided by xlibmesa-gl-dev (they have been updated and hold whatever i need)
I did not install the nvidia headers, unless the binary driver installed them itself. I have these packages installed:
normally they (the nvidia drivers) dont, so that should not be a problem.
I don't remember if nvidia puts stuff in /usr/local/include or not.
on debian it usually just puts the headers into /usr/share/doc/nvidia-glx-dev/include (but i think this is deprecated (putting headers into /usr/share/doc) and will eventually be replaced by by putting them into /usr/lib/nvidia-glx/ or sthg like that)
FYI I'm using the nvidia binary v1.0-7664
however, you can just have a look at the /usr/unclude/GL/gl.h; there are huge copyright notices at the beginning of the file that say whether it is mesa or nvidia.
I have miller's PD in /usr/ and pd-devel in /usr/local so I ran
that might be a problem
./configure with:
./configure --with-pd=/usr/bin/pd
is --with-pd=/usr/bin/pd the way to compile against miller's pd and not pd-devel?
i am wondering where this configure-flag (--with-pd) is coming from. it is nowhere used! however, even if it was used and fixed and everything i doubt whether it will be able to distinguish gracefully between installations in /usr/ and /usr/local
could you try to (temporarily) move the pddevel-include (i guess it is /usr/local/include/m_pd.h but please check beforehand) to somewhere else, and retry?
the resulting configLinux is attached.
thanks, i will give it another try...
mfg.-asdr IOhannes