
I see it in there, but i get a ToDo: error. Is it entirely unsupported, or is there a workaround that doesn't involve making a nearest-power-of-two texture?

Am Mittwoch, 6. Mai 2009 20:26:53 schrieb Jeff Heard:
I see it in there, but i get a ToDo: error. Is it entirely unsupported, or is there a workaround that doesn't involve making a nearest-power-of-two texture?
Ooops, I forgot about that one. A first look into the code suggests that the remaining bits and pieces are easy to implement. Probably I'll have some time this weekend... Cheers, S.

Also if your card support the `GL_ARB_texture_non_power_of_two' extension,
you can simply create textures of any size with the usual interface. Though
`GL_ARB_texture_rectangle' is probably more widely supported, and may
have different performance characteristics, too.
Balazs
On Wed, May 6, 2009 at 8:26 PM, Jeff Heard
I see it in there, but i get a ToDo: error. Is it entirely unsupported, or is there a workaround that doesn't involve making a nearest-power-of-two texture? _______________________________________________ HOpenGL mailing list HOpenGL@haskell.org http://www.haskell.org/mailman/listinfo/hopengl

Am Mittwoch, 6. Mai 2009 20:26:53 schrieb Jeff Heard:
I see it in there, but i get a ToDo: error. Is it entirely unsupported, or is there a workaround that doesn't involve making a nearest-power-of-two texture?
Support for GL_ARB_texture_rectangle is now in the OpenGL package repository at http://darcs.haskell.org/packages/OpenGL/. I haven't released a new version on Hackage yet, because the support for generic vertex attributes (something totally unrelated to rectangular textures) hasn't been finished yet. To explain this a bit more: I'm a bit reluctant to release an API which might immediately change in the next release. The tricky part here is that OpenGL 3.1 differentiates between 3 ways of handling integral vertex attributes: Conversion to normalized floating point, conversion to unnormalized floating point, and keeping integral values. This is not yet really reflected in the current API in the repository. Cheers, S.

Am Sonntag, 10. Mai 2009 14:00:19 schrieb Sven Panne:
[...] I haven't released a new version on Hackage yet, because the support for generic vertex attributes (something totally unrelated to rectangular textures) hasn't been finished yet.
To explain this a bit more: I'm a bit reluctant to release an API which might immediately change in the next release. The tricky part here is that OpenGL 3.1 differentiates between 3 ways of handling integral vertex attributes: Conversion to normalized floating point, conversion to unnormalized floating point, and keeping integral values. [...]
Reading the OpenGL spec very closely, one can see that the story is even more complicated: OpenGL distinguishes 3 kinds of generic vertex attributes: floating point, signed integral and unsigned integral. Integral values used for floating point attributes can be interpreted normally (i.e. value- preserving) or as normalized values. If this sounds confusing: Yes, it is! :-) Add to this a highly non-orthogonal API, and you have a perfect mess... A complete re-specification of OpenGL 4.0 (if this ever happens) instead of the agglomeration of 1-2 decades of incremental changes would be great. Nevertheless, I've tried to make this a bit clearer and orthogonal, and the result is in the OpenGL package repository. Things are not really perfect, because I want to preserve API backwards compatibility. When I find the time for a OpenGL-3.x-only package, things can be vastly simplified and would be more elegant. Nevertheless, some comments would be highly appreciated. One note: I've just seen that in my released versions, I've accidentally exported VertexAttrib without its methods and VertexAttribComponent *with* its methods. This should of course have been the other way round. :-( For compatibility reasons I keep the latter as it is, though. If there are no severe objections, I'll probably release a new version on Hackage in the next few days, including this stuff plus a few other additions and bug fixes. Cheers, S.
participants (3)
-
Balazs Komuves
-
Jeff Heard
-
Sven Panne