The Numeric Python website is http://numpy.sourceforge.net
For Numeric-21.0, run ''python setup.py install''
For Numeric-20.0, make sure to run ''setup_all.py''. In addition to
the basic Numeric modules, this command makes the FFT and other
libraries, which are very useful, but not needed by the Vision Egg.
(These extra modules are made by default in release 21.)
The PIL download area is http://www.secretlabs.com/downloads/index.htm#pil
Imaging-1.1.2 was used for development.
Unfortunately, PIL can be tricky to install.
The detailed instructions in the README are pretty good, but this package
can be a bit tricky to install.
The shell commands I use to build the Imaging packages are:
cd Imaging-1.1.2/libImaging/
./configure
make
cd ..
make -f Makefile.pre.in boot
make
If you have errors with the ''make'' step that say something like ''tk8.0''
not found, open ''Makefile'' and change ''tk8.0'' to ''tk8.3'' and ''tcl8.0''
to ''tcl8.3''. Of course this assumes you have version 8.3 of tk and tcl.
If you don't have tcl, open the ''Setup'' file and comment out the
''_imagingtk'' lines.
If you have errors with the ''make'' step that say something like ''can't
locate file: -ljpeg'', download and install those libraries or comment
out the appropriate lines in ''Setup''. I've had trouble trying to build
with those lines removed from the ''Setup'' file, so I just downloaded
and installed the libraries. These libraries are very easy to compile
and install. Just run ''./configure'' and ''make install''. Under Mac OS
X, I couldn't get a static or shared library to compile from the sources,
so I used the version that fink installed for me.
If you have to edit ''Setup'', you'll have to run ''make -f Makefile.pre.in
boot'' and ''make'' again.
Now, Imaging is compiled, and you must copy the files to Python's local
package directory. (How to find out what it is? It's usually
''/usr/lib/python2.1/site-packages'' or
''/usr/local/lib/python2.1/site-packages''.)
cp PIL.pth /usr/lib/python2.1/site-packages
mkdir /usr/lib/python2.1/site-packages/PIL
cp *.so PIL/* /usr/lib/python2.1/site-packages/PIL
PyOpenGL installation is well documented and straightforward in my
experience in linux. (Not necessarily so with other platforms!) I've
had trouble getting the GL/ARB/texture_compression.i file to compile
with the OpenGL headers that came with my nVidia card. I have a patch
that fixes this problem, if you're interested.
The Vision Egg uses pygame as a Python binding to SDL. SDL is used to
initialize an OpenGL window in a cross platform way. I have always
had good fortune with distribution installed SDL, although building
from source has always worked as well.
Once SDL is installed, installation of pygame is straightforward using
the Python distutils. Just type ''python setup.py install'' from the
pygame source directory.
Install vision egg by changing to the base visionegg directory and
execute ''python setup.py install''. You will need appropriate
privileges on your system to install.
Check your installation with the ''check-config.py'' program. Also
run this script if you run into any installation errors.