#227 PyOpenGL-accelerate and numpy 1.4 numpy_formathandler error

[jfpacker]

importing OpenGL.GL gives the error: "ValueError: numpy.dtype does not appear to be the correct type object".

To reproduce this bug, install python 2.6 (my minor version 4, ie: 2.6.4), PyOpenGL 3.0.1b2-py26-win32 and PyOpenGL-accelerate-3.0.1b2-py26-win32 and numpy 1.4. In the interpreter type:

import OpenGL
from OpenGL import GL

This gives the output:

File "C:\Python26\lib\site-packages\GLavish\library\locals.py", line 7, in <mo
dule>
from OpenGL import GL
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\GL
\__init__.py", line 2, in <module>
from OpenGL.raw.GL import *
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ra
w\GL\__init__.py", line 6, in <module>
from OpenGL.raw.GL.constants import *
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ra
w\GL\constants.py", line 7, in <module>
from OpenGL import platform, arrays
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ar
rays\__init__.py", line 22, in <module>
formathandler.FormatHandler.loadAll()
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ar
rays\formathandler.py", line 37, in loadAll
cls.loadPlugin( entrypoint )
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ar
rays\formathandler.py", line 44, in loadPlugin
plugin_class = entrypoint.load()
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\pl
ugins.py", line 14, in load
return importByName( self.import_path )
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\pl
ugins.py", line 28, in importByName
module = __import__( ".".join(moduleName), {}, {}, moduleName)
File "C:\Python26\lib\site-packages\pyopengl-3.0.1b2-py2.6-win32.egg\OpenGL\ar
rays\numpymodule.py", line 25, in <module>
from OpenGL_accelerate.numpy_formathandler import NumpyHandler
File "c:\users\jeff\appdata\local\temp\easy_install-uwdslx\PyOpenGL_accelerate
-3.0.1b2-py2.6-win32.egg.tmp\OpenGL_accelerate\numpy_formathandler.py", line 7,
in <module>
File "c:\users\jeff\appdata\local\temp\easy_install-uwdslx\PyOpenGL_accelerate
-3.0.1b2-py2.6-win32.egg.tmp\OpenGL_accelerate\numpy_formathandler.py", line 6,
in __bootstrap__
File "numpy.pxd", line 30, in OpenGL_accelerate.numpy_formathandler (src\numpy
_formathandler.c:3543)
ValueError: numpy.dtype does not appear to be the correct type object

[jfpacker]

This only occurs if you easy_install PyOpenGL-accelerate as it downloads the pre-built version win32 version. I downloaded the source and installed from source and it works.

[Mike C. Fletcher]

Hrm. Sounds like we have a problem with the numpy version changes. Argh. Seems we need to be patching setuptools to allow for checking against the numpy version. I'd hoped we'd seen the back of patching setup scripts. I'll attempt to fix it when I'm on Win32 again.

[Andrej Schoeke]

Same problem on Mac OS X 10.5 with MacPorts.