[PyOpenGL-Users] Re: Build notes...

[Mike C. F. <mcf...@ro...>]

Overall, very close to completely functional.  There were two bugs in 
PyOpenGL showing up (one related to a previously seen one), and one in 
the OpenGLContext wxContext (some key-presses aren't getting forwarded 
to the handlers).  The first bug in OpenGL is in code trying to check 
for an extension in the GLU module. The test is 
OpenGLContext/tests/nurbsobject.py:

Traceback (most recent call last):
   File "P:\OpenGLContext\tests\nurbsobject.py", line 11, in ?
     from OpenGLContext.scenegraph import shape, nurbs, material
   File "p:\OpenGLContext\scenegraph\nurbs.py", line 101, in ?
     object_space_tess.gluInitObjectSpaceTessEXT()
   File 
"C:\bin\lang\py22\Lib\site-packages\OpenGL\GLU\EXT\object_space_tess.py",
  line 13, in gluInitObjectSpaceTessEXT
     return __has_extension("GLU_EXT_object_space_tess")
TypeError: __has_extension() takes exactly 2 arguments (1 given)

Second bug is in code trying to use the gluNurbsSurface function, looks 
as though the wrapping code is generating negative indices when passed 
an array, I'm guessing this is a typed/untyped or off-by-one style error 
(typo in the GLUError as well, but I believe that's provided by the 
driver, not PyOpenGL).  You can check this with the redbook_surface (or 
redbook_trim) test(s) in OpenGLContext (or the PyOpenGL molehill demo, 
likely):

Traceback (most recent call last):
   File "p:\OpenGLContext\glutcontext.py", line 93, in OnResize
     self.triggerRedraw(1)
   File "p:\OpenGLContext\context.py", line 182, in triggerRedraw
     self.OnDraw()
   File "p:\OpenGLContext\context.py", line 132, in OnDraw
     result = mode( self ) and mode.visible
   File "p:\OpenGLContext\rendermode.py", line 16, in __call__
     self.Draw( client)
   File "p:\OpenGLContext\rendermode.py", line 33, in Draw
     self.Render( client )
   File "p:\OpenGLContext\rendermode.py", line 48, in Render
     return client.Render( self.mode )
   File "P:\OpenGLContext\tests\redbook_surface.py", line 78, in Render
     GL_MAP2_VERTEX_3
OpenGL.GLU.GLUerror: [Errno 100284] negative byte-stride encounteed

All OpenGLContext tests not mentioned above now pass.

Note: platform tested...
	Win2K, Radeon, VC6, Py2.2.1 w/ Numeric

Comments on the rest interspersed below...

Rene Dudfield wrote:
...
>  --- "Mike C. Fletcher" <mcf...@ro...> wrote: >
...
>>gluProject, seem just to be missing %name directives
>>in the wrapper), 
...
> TODO: On windows, see if this still happens?  Works ok
> here(debian/linux).
...

<snip a number of problems>
> Not sure of these problems yet.  Be good to see if it
> is happening
>  in the old version.
...
I'll check when I get some time.

> 
> I'm not sure how many people are interested in this
> library?
> Anyone interested in fixing it up?  If not, I'd be
> happy to leave
>  it as is, and note in the docs that it is
> unmaintained.

I'm interested in it in a desultory sort of way, mostly because I can 
see it being useful in such things as creating extruded text.  That 
said, I don't see it as critical to the project's success, so I'd devote 
resources elsewhere before here.

...
> I think we need some docs to describe how to build it.
>  - required libraries, how to set up system etc.
> 
> Do these exist allready?
> 
> TODO: write build docs, if not allready existing.

Sub-project for documentation added to tracker, this item added as a task.

> Also need to make the tk/tcl stuff optional.
> TODO: make build path without tcl( this
> possible/desirable? ).

Desired: definitely
Possible: I wouldn't be suprised if it's possible now, with some 
esoteric flag or another.  Problem would be finding it (then documenting 
it).  Anyway, togl is just a particular setup module, so 
worse-comes-to-worst we just yank it out and fix whatever complains ;) .

...
>>Edit the __init__ file to not import the missing
>>values as you suggested.
>>
> 
> 
> Not needed anymore.  The __numeric_* stuff comes
> through properly.

Yup, works w/out any changes.


...
> All seemed fine(except array_landscape.py) 
> Traceback (most recent call last):
>   File "array_landscape.py", line 30, in ?
>     base = landscape (2,2)
>   File "array_landscape.py", line 14, in landscape
>     base [:,:,1] = heights
> ValueError: matrices are not aligned for copy
> 
> TODO: check/fix this error.

This file is actually no longer in the project, I fixed that particular 
error a while ago, but eventually decided the whole approach was wrong 
for landscapes and just killed the experiment.

...
> This one is a framework for the other tests.
Ah, thanks.

> 
> 
>>gle/texas -- builds window, but no visible geometry
> 
> 
> TODO: see if gle/texas.py runs ok with last version. 
> see why not showing anything.

Okay, will include that in my "tests of previous version" run when I get 
the time.

...
<molehill error>
>>OpenGL.GLU.GLUerror: [Errno 100284] negative
>>byte-stride encounteed
> 
> 
> TODO: look at this and fix.  I still get same error.

See also error report for Context redbook_surface/trim, looks like the 
same error.

...
>>simple/* -- all work as expected
> 
> works.  this is a GLE test, looks like the helix one. 
> 
> TODO: see if same as GLE/helix.py one.  if so remove.
...
Looks as though the "suite" package is where Tarn was beginning work on 
a single-script test-suite?

>>tom/arraytest, poly -- weird spikes into the
>>triangle/loop, not sure 
>>why, but looks like an array storage problem, or
>>maybe a problem in the 
>>loop code
> 
> 
> looks ok here.  it's a multi colored circle.  can you
> give me a screen shot if it's still happening?
> TODO: test on Mikes system.
Bug report opened, screenshot of the test under the CVS version attached 
to it:
http://sourceforge.net/tracker/index.php?func=detail&aid=584322&group_id=5988&atid=105988

>>tom/demo -- seems to have tk menu problems (not
>>likely OpenGL related)
>>
> 
> works ok here.

Further info: bug only occurs if you close the render window before 
choosing a menu item, apparently code assumes the render window exists. 
  Not a critical error AFAIC.

...
> TODO: find someone who wants to maintain the GLE
> library.  else mark as unmaintained.
...
_______________________________________
   Mike C. Fletcher
   Why, yes, I am looking for a job...
   http://members.rogers.com/mcfletch/