Hi Robert, I would like to re-raise the issue of uniform array naming, now that it is affecting me personally :) To recap:
I have a uniform array in a shader defined as: uniform bool positions[1]; osg::Program calls glGetActiveUniform() to build a mapping of compiled uniforms by name. On some systems, the name returned is "positions" and on others it is "positions[0]". When the latter happens, a corresponding osg::Uniform named "positions" will fail since it cannot find a matching uniform location in the mapping (since it's filed under "positions[0]"). I am not convinced that this is a driver bug. It seems more like an ambiguity. The docs say that the 'uniform location for "positions" and "positions[0]" are the same; they point to the first element of the array.' (http://bit.ly/hnqbUi). I don't see anything that says the value returned from glGetActiveUniform() must be one or the other. If both names are valid and equivalent, shouldn't we account for both possibilities in osg::Program? I suppose you could just alter getUniformLocation(...) to check for both variations. Here are the past threads on the topic. The issue crops up on both ATI and NVIDIA systems. http://markmail.org/thread/5mans4howfbu2qtg http://forum.openscenegraph.org/viewtopic.php?t=1828 System: WinXP 32, ATI HD5550/10.11, OSG SVN. Thanks. Glenn Waldron : Pelican Mapping
_______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
