Change the 'count' for COMPRESSED_TEXTURE_FORMATS from "?" to "-1". This
signals to the script (glX_proto_size.py) that generates the server-side
size functions that a handcoded query is needed to determine the data size.
Also do s/FOG_COORD_SOURCE/FOG_COORD_SRC/.
When initializing the count in glx_enum::__init__, cast the local property
to int. Thanks to Python's dynamic typing, if this isn't done the type of
stored count will be a string instead of an int.
Remove the non-existant --get-alias-set option from the help text.
Update printPixelFunction (which is only used to generate code for the
server-side) to use the correct method for obtaining the dimensions of the
image parameter to a function.
Only use byteswap.h when building generated code on Linux. Elsewhere use
the (horrible!) macros from X11/misc.h.
Make sure that enums associated with the generic "Get" function get applied
to GetIntegerv, GetBooleanv, GetFloatv, and GetDoublev.
Enable GL_ARB_texture_mirrored_repeat on Savage4. Refactor the code that
sets the texture wrap mode and texture filter mode for Savage4 to the new
functions savage4_set_wrap_mode and savage4_set_filter_mode.
This was tested with texwrap and tunnel.
add _tnl_arb_vertex_program_stage to all drivers which already have _tnl_vertex_program_stage. Fixes crashes when using vertex programs. Tested on r200 only (fix by Keith Whitwell)
fix wrong fallback bit being set when ARB_vertex_program is enabled. Add missing fallback string for ARB_vertex_program. Disable ARB_vertex_program by default (just plain doesn't work). In fact, maybe it should be thrown out altogether in this branch?