[Xquartz-changes] mesa: Changes to 'refs/tags/before_upgrade_03_01_05'

Jeremy Huddleston jeremyhu at freedesktop.org
Thu Jan 1 01:04:38 PST 2015


Tag 'before_upgrade_03_01_05' created by Aapo Tahkola <aet at rasterburn.org> at 1970-01-01 00:00 -0800

Changes since the dawn of time:
Aapo Tahkola (48):
      Add radeon 9550 pci id.
      Mipmapping and other texture filters now work.
      Added fixes for filter modes and support for anisotropy filters.
      Looks like masking masking t->filter isnt needed anymore.
      I was wrong...
      This fixes it but why mipmaps still work?
      border_color causes lockups with vertex buffers. Disabling for now.
      Fixes for clamp modes.
      Add vector distance operator for vertex programs.
      Add basic sceleton for vertex programs + some other fixes
      Forgot one comment...
      Fix wrong comment.
      Cleaning up.
      Beginings of program generation. This code havent been tested nor hooked up.
      Fix for compiler warnings.
      Vertex programs work now with some restrictions. I expect arbvptorus to work
      Implemented bunch of ops.
      Implement appropriate src checks for attribs/params.
      Add a missing break and a reminder.
      Changed parts of r300_setup_routing to use WARN_ONCE.
      Changing vb code to handle inputs correctly.
      Removing some trash and disabling fog fallback as it seem its causing more harm than good. Tuxracer should now work at least with vbs. Fix for vb colorbuffer clear bug is underway...
      To cut the long and boring story short, polygon offsets should now work. vb mode colorbuffer clears are still mysticly broken by offsets... Maybe we need to merge zbs and unk42B4 together?
      Implemented checks that prevent r300 from locking up when bad number of verts are given. Also tweaked WARN_ONCE a bit to handle va-args.
      fog + immediate mode = bad things
      Better fix to get tuxracer working.
      BIG FAT NOTE: Theres a nasty bug somewhere thats causing vb color buffer clears and other things not to work. This bug can be triggered by extending struct r300_hw_state by two struct r300_state_atom's from its current size. Everything zbs and unk42B4 related is now covered with HAVE_ZBS and GA ifdefs. Who wants to fix it? Not i. :)
      Fix the bug and get zbias back into shape. There might still be some problems with initial zbias...
      Typo..
      Enabled user-defined point size, enabled zbiases for rest of the primitives and fixed bad vertex number checks triggered by nurb.
      Adding support for selecting line width. Stipple patterns still cause lots of trouble.
      Turns out unknown5 was border color. texwrap looks good now.
      Support for PolygonMode. Not too far from working even though i almost gave up once already.
      Removing some regs that are wrong and adding some comments to r300_reg.h .
      Updating master copy of r300_reg.h and turning immediate mode back on
      Added kindly note about single buffer mode.
      Support for blend color(not tested), "proper line width/point size" clamps and fix for glxinfo problem.
      Support for idx buffers. Leaving it on by default as it doesnt seem to cause any lock ups nor other issues. Tests with one object using elts should pass. Introducing more than one object will cause indices to mix up as far as i can see. DRM update is needed for this code to work\!
      magic_1 handling...
      Switching to start_index32_packet.
      We dont need to call r300EmitAOS if we arent going to draw anything(r300_get_num_verts returns 0).
      Fix for buffer overrun caused by ALLOC_STATE not having args surrounded by parenthesis. Can you see it? HINT: Anything texture related should now work slightly better. And yes it took me several hours to find it.
      Disabling some fallbacks as they cause misc programs not to start and adding some comments.
      Minor fix to border and blend color.
      Adding support for front buffer mode, removing some code we dont really need and some new debugging stuff.
      New elt buffer code should be fairly stable.
      Another nasty struct size bug found.
      fix for 'nasty bug' and some sanity checks to avoid buffer overruns. Bumping VSF_MAX_FRAGMENT_LENGTH as it seems a bit low otherwise.

Adam Jackson (40):
      Skeletal fallback-only DRI driver.  Initial checkin, not quite working yet.
      Fix type conflict.
      use new interface for tdfx (phase 2)
      silence a warning
      _really_ fix parameters to StoreImage calls.
      Bug #1588: abort if no palette format matches. (Egbert Eich, Stefan Dirsch)
      Bug #1682: Mesa core code that gets linked into DRI drivers should never call
      Import sarea.h from Xorg, needed for client GLX code.
      Import the GLX client side library, formerly from xc/lib/GL/glx.  Build it
      Bug #1679: Link DRI drivers against DRI_LIB_DEPS, not GL_LIB_DEPS.
      Add DRI_LIB_DEPS for the DRI drivers to link against.  Remove expat from the
      Add glx/x11 to the DRI configs, and change the Solo configs to build glx/mini.
      remove a -Werror from testing.
      Bug #1713: Some rare libGL's have __glXFindDRIScreen defined but do not
      libGL needs libXxf86vm.a.
      static char * -> static char []
      constify a few arrays.
      make render_quads_verts call EMIT_PRIM with the arguments in the right order,
      brown paper bag, r100 can't do hardware quads.
      Sync with Xorg head: mingw build fixes from Alexander Gottwald.
      Fix up glx/x11 to work when built with -fvisibility=hidden.
      fxt1_decode_1() needs to be non-static for the tdfx and glide drivers.
      Mark the XF86DRI client-side extension API as PUBLIC.
      _glapi_* is effectively PUBLIC API for the DRI drivers.
      Mark __driCreateNewScreen PUBLIC
      Also PUBLICize the config option table
      Import s3virge and trident drivers.  Not functional yet; no Makefile, no DRI-aware DDX.
      Get trident building and -Werror clean.
      Get s3v building with a minimum of warnings.
      Build s3v and trident by default too.
      Bug #1859: Initialize 'size' before first use in intelTryReadPixels()
      Unused variable cleanup.
      Fix read-from-uninitialized in s3vMakeCurrent()
      linux-dri-x86 build fix from Andreas Stenglein.
      Parallel build fix.  Multiword variables on the target side of a : get
      Enable libGL to be built with DRI_NEW_INTERFACE_ONLY (but don't do it yet).
      silence makedepend warnings about non-portable whitespace
      HAVE_ALIAS was never defined anywhere, add some logic to turn it on
      Bug #2428: #ifdef GLX_DIRECT_RENDERING in DRI drivers is pointless.
      Bug #2477: Make GLU/GLw/glut build properly even with -fvisibility=hidden.

Alan Hourihane (117):
      fix cast
      fix glxinfo to query current display.
      add FINISH function at end of each render_* function
      Add quad and quad_strip
      fix use of projected vertices with notex vertices when DO_PTEX isn't set.
      remove a HAVE_ELTS that I'd stuck before.
      Don't reset _tnl_CurrentInput when in DestroyContext.
      Correct check for noop array state change (Raystonn)
      Test for NULL pointer for LoadMatrix(), MultMatrix() and
      glutSwapBuffers implicitly does a glFinish when swapping back to the front
      remove test code
      Check that ActiveTextureARB doesn't overflow MaxTextureUnits
      fix for framebuffer width and height when window hasn't been realized.
      fix fragment_program for Windows
      cast GLfloat
      mkdir the server directory, as using cvs with -P would prune an empty
      add the i810 driver - no kernel driver yet
      add the i830 driver - no kernel driver yet
      fix cut & paste errors
      add the SiS driver - no kernel driver yet
      add gamma driver - no kernel driver yet
      char -> GLubyte
      prevent from optimizing out by some compilers (from XFree86 CVS)
      bring over glcore.h changes from DRI trunk
      add tdfx DRI driver
      build fixes
      add SPARC ffb DRI driver
      build fixes
      build fix
      build fixes
      remove duplicate declaration
      #define GLAPIENTRYP GLAPIENTRY *
      check if GLAPIENTRYP is already set
      build fixes
      build fixes
      build fixes
      build fix
      merge latest DRI sis driver changes from the DRI trunk
      use _swrast_CopyPixels for XFree86
      disable some DEBUG
      bring in common from DRI trunk
      bring over latest mga DRI driver from DRI trunk
      remove duplicate declarations
      fix compiler warning
      merge i810 DRI driver from DRI trunk
      merge r128 DRI driver from DRI trunk
      xmlconfig fixes for the new options
      xmlconfig fixes
      fix some _SOLO build problems
      fix _SOLO build problem
      fix a _SOLO build problem
      fix _SOLO build problem
      merge i830 DRI driver from DRI trunk
      sync some more of the r200 driver from the DRI trunk
      merge some more of the radeon DRI driver from the DRI trunk
      add SETPARAM ioctl
      reverse Keith's 'XFree86Server' glue.
      kill a warning
      build in SOLO environment
      some changes for SOLO build
      turn on tdfx in SOLO build
      fix bogus assertion that checked for an empty texture heap list before
      fix typo
      fix from DRI trunk
      fix for C++
      fix printf warnings
      remove non-portable whitespace
      build fixes
      make note about needing to fix the drm driver to enable SAVAGE_CMD_DMA
      build fix for VMS
      fix the usage of GLX_MESA_allocate_memory
      disable under _SOLO build
      add SciTech's GLDirect driver for Windows.
      bring over build fixes from stable branch
      remove DitherValues
      If we can't allocate the rowimage - bail
      re-enable rect texture
      fix for older gcc's (from freedesktop bug id #419)
      Fix some warnings
      reverse some debug that slipped through
      silence warnings
      silence warning
      Check for some header defines before redefining functions. Silences warnings.
      silence warnings
      Use union type to avoid strict aliasing problems.
      uint*t -> u_int*t changes
      fix warning
      uint*t -> u_int*t
      Bring VIA driver up-to-date with regard to drm_*_t changes and remove
      fix build problem
      Only try to free the front buffer when it's a pbuffer.
      support 720x480 and 960x540 modes
      Add Intel i915GM support, and these extensions.
      Fallback on PolygonStipple for CLE266 hardware.
      Fix usage of texture units, when TEX1 is enabled, but not TEX0.
      fallback on polygon.stippleflag (thanks Keith)
      Fix ColorMask
      uncomment
      Fix the calculation of the alpha reference value which negates the
      fix a typo
      Fix multitexturing.
      build fixes and enable agp dma
      Fixes lockups initializing AGP DMA.
      Check for deletion of currently bound context
      Use the faster span read/write template for 16bpp
      As we fallback for polygon stipple on the CLE266 for now, ensure we
      Fix glean scissor test
      revert some makecurrent code
      remove some dead texture code
      16bpp doesn't support masked clears, so fallback when they're enabled.
      Get PolygonStipple working.. (or seemingly too). It seems to do the right
      silence a warning
      Enable line stippling.
      Fix tristrips (which fixes coloredTexPerf2 & coloredLitPerf2 glean tests)
      remove some debug
      Enclose passed macro values in brackets to ensure correct read/write span
      silence the 'using AGP/PCI' string unless VIA_DEBUG is used.

Alex Deucher (1):
      Apply Eric's stencil wrap patch (Eric Anholt)

Andre Werthmann (3):
      checked in the first new PIII vertex-transformation code
      added more ASM_SRCS
      added the katmai_transform_points4_* functions

Ben Skeggs (7):
      Added GL_POLYGON primitive. Single primitive between glBegin()/glEnd() now works.
      Port of r200 VB code.  VB path should be as fast as the immediate path now, tuxracer should render correctly.  Immediate path was left enabled.
      Enabled vertex buffer mode as default.
      Share routing code between immd/vb.  Fix breakage of vertex programs in vb mode.
      Don't enable fog.  It's broken
      Make immediate code do a LOAD_VBPNTR again
      oops.

Brian Paul (3674):
      initial rev
      Initial revision
      added Daryll, David, and Steve's ideas
      added MESA_HPCR_CLEAR env var info
      added a few URLS
      added info about overlay planes
      Initial revision
      updates from Vik
      Initial revision
      added note about new driver for BeOS R4
      removed old driver info
      Initial revision
      added Special Features section
      initial rev
      initial rev
      initial rev
      added GL_LINEAR 3D texture bug
      added CVS Id tag
      updates
      initial check-in
      added bug fix note for wide points and lines
      updated by Alexander Mai
      updated 3.1 beta 2 data
      updated by Alexander Mai
      Updated OpenStep/NeXT support (pete at ohm.york.ac.uk)
      Ted's 3.1b2 update
      added 3.1 beta 3 info
      initial rev
      clarified what this file covers
      optimized glAccum function
      testing
      glIsEnabled(GL_TEXTURE_3D) was wrong
      minor change for irix-debug
      initial check-in (post-crash)
      initial check-in (post crash)
      fixed alpha channel problem in gl_blend_pixels()
      added a few more bug fixes
      added vertices.c
      removed TexCoordUnit from GLcontext, use Array.ActiveTexture instead
      minor changes. always call compactify_arrays()
      moved gl_GetString() into get.c
      moved gl_GetPointerv() into get.h
      changed C++ comments to C style
      added GL_EXT_get_proc_address
      added GL_EXT_get_proc_address
      added GLX_EXT_get_proc_address
      added GLU_EXT_get_proc_address
      include extensions.h
      GL_EXT_get_proc_address clean-up
      added 1.2 function to gluGetProcAddressEXT()
      clean-up of glXGetProcAddressEXT()
      fixed typo in GetProcAddress
      added Eero's multitexture function patch
      added gluUnProject4()
      added gluCheckExtension()
      fixed pixel packing/unpacking code in gluBuild2DMipmaps()
      added linux-386-pthread-shared-glide
      removed unused GLcontext arg from a few functions
      added TINY version symbol
      added third, tiny version number to arguments
      added third, tiny version number to mklib scripts
      added a few more files to the tar list
      new version number symbols
      new MESA_ version number symbols
      3.1 beta 3 info
      removed mention of libMesaGL
      minor library name changes
      misc changes to tar file list
      bumped version strings to Mesa 3.1
      changed version string to return 1.2
      added GLX_MESA_set_3dfx_mode.  clean-up of glXGetProcAddress
      initial check-in
      added glxinfo
      changed archive filenames to beta3
      added GL_EXT_get_proc_address
      GL_EXT_get_proc_address clean-up
      clean-up of GLU_EXT_get_proc_address
      GLU_EXT_get_proc_address clean-up
      added some casts in gluGetProcAddressEXT()
      initial check-in
      updated for GL instead of MesaGL
      gluGetProcAddressEXT change for C++ / BeOS
      version symbol changes
      link with -lGLU -lGL instead of -lMesaGLU -lMesaGL
      added src-glut.beos back in demo files
      link with -lGL instead of -lMesaGL
      fixed typo
      pad GLU_TINY to 6 digits
      pad GL_TINY to 6 digits
      glGetProcAddressEXT changes to accomodate Win32 and non-Win32
      silenced some warnings
      added a bunch of missing files to LIB_FILES
      disabled GL_EXT_get_proc_address for the time being
      fixed reference count bug in save_TexImage3DEXT()
      texture units now share default texture objects
      added memory macros
      now using GL_MALLOC, GL_FREE
      fixed dumb typo
      now using GL_MALLOC, GL_FREE
      now using GL_MALLOC, GL_FREE
      use texture objects now
      removed GL_ prefix from memory macros
      now using MALLOC, CALLOC, FREE, etc macros
      fixed an RGB vs CI bug in overlay visual selection
      replaced glGetMinMax() with glGetMinmax()
      added sco5 config
      removed GL_EXT_get_proc_address
      removed GL_EXT_get_proc_address
      added missing GL_ARB_imaging color table functions
      more ARB_imaging colortable functions
      added glGet cases for vertex array enable state
      negative offsets to TexSubImage w/ texture borders caused memory corruption
      GL_MAX_ELEMENTS_VERTICES/INDICES enums had wrong values
      added more format/type error checking code
      clean-up in gl_LightModelfv()
      more image type error checking in draw_stencil_pixels()
      fixed a number of error-test bugs
      enable lighting even if no light sources are turned on
      fixed divTable precision problem in GL_RETURN case
      clean-up of optimized GL_RETURN code
      silenced uninitialized variable warnings
      silenced ininitialized var warnings
      silenced some uninitialized var warnings
      updated some comments/docs
      added -info command line option
      added f key to toggle front/back drawing
      applied Jonn Carmack's patch for faster glTexSubImage2D() in Quake
      minor clean-up of linux-elf-debug target
      define GLUAPI, GLAPIENTRY and GLCALLBACK to nothing if not using Mesa's gl.h
      removed revision history info
      added optimized GL_RGB, GL_UNSIGNED_BYTE case to gl_pack_rgba_span)
      fixed byteswapping bug in gl_pack_rgba_span()
      initial rev
      added gloss demo
      more comments, removed unneeded glDepthFunc call
      fixed stupid error in GL_RGA case in gl_pack_rgba_span()
      better texture matrix, better end-cap reflections on cylinder
      removed gluGetProcAddressEXT()
      removed glXGetProcAddressEXT
      removed cvs history logs and old comments
      disabled gluGetProcAddressEXT
      disabled glXGetProcAddress code
      added -lm to XLIBS for all configs
      removed -lm from link command
      minor changes to Usage() function
      added a few more scale/bias/lookup functions (for future use)
      pixel pack/unpack ImageHeight and SkipImages was not initialized
      fixed a typo in a comment
      added linux-glide-debug config
      new texture image, cleaned-up code
      new texture image processing
      added another fast-path in _mesa_unpack_ubyte_color_spa(), more comments
      include assert.h with rest of system headers
      clean-up to reduce MSVC warnings
      new xfont demo
      added xfont demo
      removed printf's
      clean-up of header includes (Daryll)
      flush pb after each line segment - for conformance
      changes to silence MSVC warnings
      header includes clean-up (Daryll)
      replace GLint with GLsizei in a gluScaleImage, gluBuild1/2DMipmaps()
      fixed typo: replaced |= with != in conditional
      first big check-in of new Mesa 3.3 code
      initial changes for Mesa 3.3
      added missing braces in array initializations
      don't include strings.h or stdio.h
      updated for new texture color palette struct
      removed redundant extension prototypes
      ColorTableEXT clean-up
      changed a warning string
      minor clean-ups
      added some braces to silence gcc warnings
      added gl_extension_is_enabled()
      added GL_EXT_texture_env_add extension
      initial check-in
      added glxdpyinfo
      restored the gl_Viewport function
      moved assembly-related macros from types.h to glheader.h
      reordered some functions
      replaced _GLAPI_EXT_color_table with _GLAPI_EXT_paletted_texture
      replaced EXT_color_table with EXT_paletted_texture
      changed multi-include guard symbol
      update/merge with Daryll's X server changes
      fixed mistake in merge regarding texture palettes
      include mem.h
      texture palette update
      fixed several texture state bugs found with objbug.c program
      make clean removes .lo .la and .libs/
      fixed some glitches (Eero Pajarre)
      test visinfo, not vishandle in save_glx_visual (Wolfram Gloger)
      silence an uninitialized var warning
      removed unused fxTMNewTMFreeNode decl
      fixed FX_grSstControl(), now returns FxBool
      reordered the extensions and numbered them
      two fixes in gl_DrawArrays (Andree Borrman)
      replace encounteed with encountered
      updates for Mesa 3.3
      updates for thread safety
      renamed struct gl_palette as struct gl_color_table
      removed redundant fxCloseHardware() decl
      added pragma 4615 (Eero), added copyright
      fixed shade model bug (A.Borrmann)
      added SET/CLEAR/TEST_BITS macros, misc clean-up
      fixed potential bug in _mesa_ShadeModel()
      bumped GLX version to 1.2, added glXGetCurrentDisplay()
      added glXGetCurrentDisplay() for GLX 1.2
      changed comment for WriteRGBSpan()
      removed GLU_EXT_get_proc_address from ext strings
      clean up of version handling, removed GLX_EXT_get_proc_address string
      added _mesa_clip_pixelrect()
      added GLX 1.3 functions
      added GLX 1.3 tokens and functions
      added GLU 1.3 functions
      renamed near and far arguments to gluUnProject4()
      added a few more GLX 1.3 tokens
      added XMesaMakeCurrent2(), XMesaGetCurrentReadBuffer(), bumped version to 3.3
      bumped version to 3.3
      removed version logging
      added support for separate read/draw buffers per context
      updated with Mesa 3.3 info
      updated for Mesa 3.3
      replaced Driver.SetBuffer() with SetDrawBuffer() and SetReadBuffer()
      implemented glXMakeContextCurrent() and glXGetCurrentReadDrawable()
      bunch of changes for separate read buffer feature
      initial rev
      added wincopy demo
      include glapi.h
      initial rev
      now using glapitemp.h to generate entrypoints
      include glapitable.h
      generate Z values when fog enabled (Andree Borrmann)
      updates for 3.1 and later
      removed Mesa-isms
      added _GLAPI_EXTRA_SLOTS symbol
      lots of work on extension management
      added _mesa_init_no_op_table()
      added call to _mesa_init_no_op_table()
      move dispatch table size check into _mesa_init_no_op_table()
      initial rev
      all new dispatch system
      changes for new GLX dispatcher
      obsolete
      removed realglx.c
      removed include fakeglx.h
      updated comments, removed obsolete functions
      added missing functions in GLX_functions[] array
      print debug info in gl_make_current() if MESA_INFO env var set
      fxInitPixelTables() change for Voodoo 1/2 vs 3
      gl_BindTexture was broken when target==GL_TEXTURE_3D
      added WM_SYSKEYDOWN case in __wglMonitor() (Eero Pajarre)
      test for null newCtx before newCtx->FirstTimeCurrent
      added glCopyPixels bug fix
      minor stencil-related changes
      Clearing stencil buffer now observes the stencil write mask
      removed gl_get_proc_address()
      added glXGetProcAddressARB()
      code clean-up
      added stencil buffer functions
      updated for hardware stencil support
      applied Daryll's patches
      changes in hardware depth buffer support
      pass new args to gl_create_framebuffer()
      applied Daryll's patches
      pass new args to gl_create_framebuffer
      new depth buffer functions
      added GL_ARB_transpose_matrix
      implemented GL_ARB_tranpose_matrix
      added GLX_ARB_get_proc_address to extension list
      added GLX_ARB_get_proc_address
      bitmaps were vertically shifted by one pixel
      removed unneeded code in gluBuild1DMipmaps()
      added many more extension entry points
      removed _GLAPI_ extension testing
      replaced 0 with 0.0 in sin, cos calls
      moved #define to column 0
      made a few globals static
      just include Makefile.X11
      initial rev
      new thread code
      added _glapi_get_proc_name()
      added a cast to malloc call
      initial rev
      added more extension functions
      added more functions
      new threading, more functions added
      minor change
      initial rev
      thread support now works
      minor clean-up
      don't free thread data in pthread_key_create()
      more tweaking of thread support
      fixed thread problems
      removed old thread code
      added current context functions, made ThreadSafe public
      further simplification of thread-related code, misc clean-up
      replaced mthreads.c with glthread.c
      updated with latest source files
      updated for new thread support in Mesa 3.3
      print thread safety info if MESA_INFO set
      optimized GET_IMMEDIATE for thread safety
      added missing glColorTableParameter defines
      start search at 1 in HashFindFreeKeyBlock()
      removed _glapi_ThreadSafe
      clean-up of Linux target configs
      misc updates
      minor clean-up in _mesa_unpack_ubyte_color_span()
      new Clear function mask
      updated Clear function for new mask bits
      return GL_FALSE in fxDDColorMask
      assigned official GL_ARB_transpose_matrix enums
      fixed GetString template, added GL_SGI_color_table templates
      removed cpp conditionals from _mesa_init_exec_table()
      updated version string
      added BitsPerPixel to XMesaVisual struct
      added glHintPGI template
      improved robustness of extension handling in static_functions[] array
      added missing glHintPGI() function
      minor updates
      fixed prototype typo in glHintPGI ( )
      removed DISPATCH_SETUP, added RETURN_DISPATCH
      added amesa.h to tar file list
      added printf-message to dispatch templates
      fixed missing slice problem in gluPartialDisk()
      another fix for gluPartialDisk() in GL_LINE mode
      made some functions static
      minor clean-up
      commented out unused function
      commented out gl_reinit_light_attrib()
      clean up of eval_coord1/2f()
      include dispatch.h
      init a few vars to silence compiler warnings
      added gl_register_config_var() proto
      added _mesa_ArrayElement() proto
      added -Wmissing-prototypes and enabled pthreads for linux debug configs
      applied Daryll's patches for GLX_DIRECT_RENDERING
      added fxMesaContext arg to fxInitPixelTables()
      added gl_initialize_context_data(), gl_free_context_data(). code clean-up
      GLcontext is now nested in OSMesaContext
      updated for Mesa 3.3
      first check-in
      minor change for DEBUG builds
      use nplanes instead of bitsPerRGBValue in GET_BITS_PER_PIXEL macro
      Z values were wrong in selection mode (Holger Waechtler), misc clean-up
      added missing POINT/LINE RANGE/GRANULARITY tokens
      added copyright info
      removed X86 files from linux debug configs
      renamed _mesa_Dispatch to _glapi_Dispatch, misc clean-up
      added _glapi_Dispatch
      disabled X86-optimized dispatch when using XF86DRI (temporary)
      added X86-optimized message for MESA_INFO
      added do /while to multi-line macros
      minor clean-up
      renamed CURRENT_INPUT to _mesa_CurrentInput
      generic_noop() now returns 0
      replaced some spaces with a tab
      updated with latest DRI tree changes
      updated SVGA sources from Mesa 3.2
      updated for Mesa 3.2
      added new SVGA driver files
      added #ifdef SVGA, work-around asm code problem
      prefixed hash functions with _mesa_
      replaced HashLookup() with _mesa_HashLookup()
      changed VB_MAX, no longer dependant on FX symbol
      updated for Mesa 3.3
      fixed VB->IndexPtr==NULL bug (Richard Guenther)
      only print info if MESA_DEBUG is set
      added info about assembly language optimizations
      updated RGB2BGR function
      added WIN32 and macos files to tar archive
      include conf.h if HAVE_CONFIG_H defined
      all new version of glxinfo
      updated version string to Mesa 3.3
      minor changes regarding transparency and multisample
      added improved glxinfo item
      initial check-in
      added texdown program
      added texdown info
      updated comments
      added default template macros
      enable/disable no-context warnings depending on MESA_DEBUG
      moved entrypoints here from glapi.c
      removed actual GL entrypoints (glapitemp.h)
      renamed _glapi_CurrentContext to _glapi_Context
      added -I. to .S.o rule
      include glapioffset.h instead of ../glapioffsets.h
      added Josh's dynamic display assembly generator
      fixed protos for glTexImage3DEXT and GetDetailTexFuncSGIS
      added more function templates, changed message format
      changed message format, added GL_EXT_texture3d
      removed unneeded \n chars
      added save_TexImage3DEXT()
      added _mesa_TexImage3DEXT()
      added Mutex macros
      added mutexes for thread safety
      defined FLT_MIN if undefined
      added more message strings
      added normal, vertex, viewport message strings
      patch for XFree86 building
      limit Drift to [0, 1]
      changed > to >=
      replace offset with functionOffset in generate_entrypoint()
      functions from old glmisc.c file
      functions from context.c
      removed glmisc.c, added buffers.c, hint.c, state.c
      moved state functions to state.c
      updated includes for source file reorg
      moved exec dispatch init to state.c
      updated source files
      include buffers.h
      added comments
      don't include dispatch.h
      try both single and double-buffered visuals
      changed prefix to _mesa_ for a few accum functions
      added check for IntegerAccumScaler > 0 in GL_RETURN case
      replaced gl_ prefix with _mesa_ prefix
      replaced gl_ prefix with _mesa_ prefix on blend funcs
      replaced gl_ prefix with _mesa_ prefix on depth funcs
      changed _gl prefix to _mesa_ on fog functions
      minor clean-ups
      added Const.SubPixelBits
      added point/line size limits to gl_constants struct
      added SUB_PIXEL_BITS
      pass IntFormat, not Format to fxTexGetFormat
      added support for GL_RGB5_A1 internal texture format (Eero Pajarre)
      GL_ARRAY_ELEMENT_LOCK_FIRST/COUNT_SGI changed to _EXT suffix
      disabled locked arrays in glDrawRangeElements()
      added missing glGet cases for GL_EXT_compiled_vertex_array
      updated with new info
      clearing depth buffer failed when clearing front color buffer
      in fxSelectSingleTMUSrc_NoLock(), added check for haveTwoTMUs (Neal Tringham)
      added a few missing ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH() calls
      glDrawPixels w/ format=GL_COLOR_INDEX was broken
      updated version info
      updated fxDDGetString() for DRI
      removed fxDDBufferSize proto
      udpated version/copyright info
      added OpenStep to LIB_FILES
      added GLCALLBACKPCAST for Cygnus (Sven Panne)
      initial checkin
      include readtex.h
      added GLCALLBACKPCAST for Cygnus (Sven Panne)
      added support for XFree86 threads. misc clean-up
      changed thread TSD initialization
      THREADS now defined by glthread.h, added more prototypes
      added tess_clip.c
      enable thread safety on Linux configs, removed -DTHREADS flag
      clean-up
      added code for all GL_MESA_window_pos functions
      fixed WIN32_THREADS error
      updated with MESA_FX_NO_SIGNALS info
      added MESA_FX_NO_SIGNALS env var (Michael Vance)
      fixed reference count but in DeleteTextures()
      always include <float.h>
      clean-up
      now using dynamically allocated api dispatch tables
      added more error checking to _glapi_add_entrypoint()
      always include thread.h
      initialize fxMesa->verbose with MESA_FX_INFO env var
      new library dependency vars
      updated with changes from DRI project CVS
      update fog near/far bug fix
      replaced renderer_string() with get_string() func
      replaced gl_ggiRendererString() with gl_ggiGetString()
      removed driver RendererString() and ExtensionString() funcs
      added CHAN_MAX, CHAN_MAXF
      fixed GL_SRC_ALPHA_SATURATE bug, lots clean-up
      fixed bug in  GL_UNSIGNED_INT_10_10_10_2 unpacking
      fixed minor bug in copytexsubimage_error_check()
      fixed off by one error in matrix stack depths (Eero Pajarre)
      added aatriangle.c source
      disabled an fprintf in FX_grSstQueryHardware()
      fixed bug in 24bpp ximage pixel addressing
      initial check-in
      sort the offsets by number
      initial revision
      added some missing tokens, fixed some errors
      added glext.h header item
      added a bunch of extensions
      include GL/glext.h
      fixed a typo in GL_MESA_resize_buffers
      fixed a few glitches
      changed a GLuint to GLint
      GL_EXT_blend_color is now DEFAULT_ON
      fixed problem in print_extension_list() when ext ended in a space
      changed glXCopyContext()'s mask to unsigned long
      added more extension entrypoints
      added missing backslash on line 483
      use quotes for gl.h and glext.h includes
      added some error checking
      bunch of extension clean-up
      bunch of dispatch and extension-related changes
      added more entrypoints
      now generated with Python script
      now using SGI-specified dispatch order
      new fxDDClear(), fxDDDrawBitmp(), example extension registration code
      added note about double buffering for in-window rendering
      fixed front/back buffer stride problem for XF86DRI
      changed glXCopyContext's mask to unsigned long
      changed glXCopyContext mask back to GLuint
      added XMesaUnbindContext (for DRI)
      removed the GET_BITS_PER_PIXEL macro
      inserted a tab
      added GL_HAS_GLEXT
      fixed bug when using display lists
      added all the PFNGL...PROC typedefs
      patched for NetBSD support (Allen Briggs)
      added a bit of GLAPIENTRY, APIENTRY setup
      changed glXCopyContext() mask to unsigned long, per GLX spec
      added GL_BGR and GL_BGRA support
      added proxy texture targets to GetColorTableParameter (Steven Fuller)
      test for GL 1.2
      added GL_GLEXT_VERSION_EXT
      test glDraw/Read/CopyPixels()
      added readpix program
      minor reformatting
      inserted static const in make_null_texture()
      simplification of GLNAME macro setup (David Dawes)
      minor clean-up
      changes to _WIN32 test and APIENTRY test
      removed GLX_VERSION_1_3 for now
      added device driver ReadPixels and CopyPixels functions
      moved device driver DrawPixels call
      save ColorTable and ColorSubTable commands
      runtime selectable depth buffer depth
      clean-up warnings
      depth buffer change, BGR/RGB setup change
      added LineWidth device driver function
      removed obsolete logicop function
      fixed off-by-one errors in matrix stack setup/cleanup
      initial check-in
      added geartrain
      added demos/geartrain
      now save Targa image.  misc clean-up
      added target parameter to Driver.TexEnv()
      made gl_make_fog_coords() static
      added GL_EXT_texture_lod_bias extension (Michael Vance)
      added GL_HP_occlusion_test, GL_EXT_texture_lod_bias
      added GL_EXT_texture_lod_bias
      implemented FX_grSstPerfStats()
      minor clean-up
      removed Driver.UseGlobalTexturePalette()
      glPopAttrib() now calls Driver's ClearColor, ClearIndex and ColorMask functions
      initial implementation of GL_HP_occlusion_test extension
      gave temp values to GL_HP_occlusion_test tokens
      added final GL_HP_occlusion_test values
      fixed polygon stipple packing/unpacking
      removed old comment
      removed old Depth buffer pointer var
      added missing error checks
      silence some warnings
      colors were computed incorrectly (Hans Nelles)
      removed MGL files
      added glext.h to tar file list
      fixed compiler warnings on BeOS R4
      updated for Mesa 3.3
      removed debug printf
      removed GL_FXMESA_global_texture_lod_bias extension
      added aa triangles
      added device driver functions for CopyTexImage, CopyTexSubImage, GetTexImage
      patched for Cygwin (Sven Panne)
      Cygwin changes, removed two ARB extensions, now in glext.h
      removed test for MESA symbol
      added more extensions, now at version 3
      initial check-in
      added lodbias demo
      added new_texture_image()
      added new TexImage and TexSubImage functions
      rearranged order of some functions
      more reorganization and clean-up
      added _mesa_image_row_stride()
      clean-up of texture image construction
      changed imaging functions to _mesa_ prefix
      made _mesa_native_packing const
      removed retainInternalCopy from TexSubImage functions
      added more locking
      removed a few const keywords because of mutex calls
      fixed Cygwin patch typos
      lots of changes for single-copy texture image DD interface
      updated renderer string version to 20000322
      tweaked the animation rate
      converted from GL_EXT_texture3D to GL 1.2
      changed TDA to [4] for glMaterialfv() calls
      added code to die on FP exceptions if DEBEG env var set to FP
      added Daryll's patches for texture alignment
      minor tweaks
      added instructions to build widgets-sgi
      more clean-up and rearranging
      replaced decode_internal_format() with _mesa_base_tex_format()
      removed some unpopular extensions (now in glext.h)
      added 3.3 To Do items
      replaced HaveTexObj with UseTexObj
      added some comments and new texture image functions
      added new texture image functions, minor clean-up
      changed some Linux -O2 flags to -O3
      added benchmarking
      version 4, all extensions up to 197, added some missing enum values
      initial check-in
      added texutil.c source
      fixed a Driver.TexImage call typo, misc clean-up
      added missing glEnd() call (Tim Beckmann)
      put _mesa prefix on some functions
      added _mesa_unconvert_teximage()
      removed tfxMipMapLevel's used and translate fields, initial teximage clean-up
      added optimized read/drawpixels for 16bpp TrueColor
      version 5: added texture compression extension
      lots of changes
      fixed some buffer clear bugs
      added missing GL_INTENSITY case in _mesa_components_in_format()
      added error checking on file open
      hit space bar to rerun benchmark
      changes for single-copy textures
      lots of new work
      new texture image download code
      renamed mesaStrDup() to str_dup()
      removed GL_HP_occlustion_test code, added -exit option
      fixed subtexture address calculation bug
      fixed typos in generate_lookup_tables()
      improved PrintTexture debug function
      optimized glReadPixels()
      tweaks to allow different read formats/types
      added GLX_EXT_visual_rating
      added XMesaCreateVisual2()
      separate R/G/B/A accum buffer size fields
      added _mesa_create/destroy_visual()
      updates for separate R/G/B/A accum buffer sizes
      minor tweak to GLX_VISUAL_CAVEAT_EXT query
      added rest of GLX_EXT_visual_rating extension
      added GLX_EXT_visual_rating
      added more debugging code
      misc clean-up
      fixed Voodoo1/2/ stride problem in fxDDReadPixels()
      clean-up of info messages
      fixed typo in GLX_SLOW_VISUAL_EXT
      added glu_tar and glut_tar targets, misc clean-up
      mask/disable stencil/accum clears in fxDDClear()
      fixed mistake in glXGetConfig(GLX_VISUAL_CAVEAT_EXT)
      added support for GLX_EXT_visual_rating
      changes for GL_HP_occlusion_test
      initial check-in
      added occlude demo
      added GL_HP_occlusion_test items
      minor clean-ups
      use BCOPY macro on FreeBSD
      more GL_HP_occlusion_test work
      updated for Mesa 3.3
      added copyright info
      changed a comment
      added version number
      minor clean-ups, renaming
      updated to version 6
      added missing braces in initializers
      fixed compiler warnigns
      removed unused vars
      added extern for gl_fog_coord_stage
      added some casts
      new arguments to XMesaCreateVisual()
      silenced a bunch of warnings
      silenced warnings
      initial histogram and min/max work
      new read_R5G5B5_span() function
      silenced a warning
      version 7 check-in
      fixed a typo
      added GL_SGIX/SGIS_pixel_texture extension
      added pixeltex.c
      initial check-in
      added pixeltex demo
      added pixel texture extension info
      added GL_SGIX/SGIS_pixel_texture
      initial check-in
      added GL_SGI_color_matrix extension
      added GL_EXT_color_matrix extension
      removed a printf
      fixed GL_PGI_misc_hints identifiers
      fixed visual selection and reporting results
      added missing 1.3 token values, added glXGetFBConfigs(), fixed glXChooseFBConfig()
      removed bogus GL_ALLOW_DRAW_SPN_HINT_PGI token
      added glXGetFBConfigs(), fixed glXChooseFBConfig()
      added some more functions
      fixed stride problem in read_R5G6B5_span
      initial work on GL_SGI_color_table extension
      clip depth/stencil spans to window bounds
      fixed bug in XF86 writeRegionClipped()
      silenced warnings in FX_grSstPerfStats()
      fixed duplicate variable problem
      renamed stencil functions
      renamed alpha buffer functions
      more GL_SGI_color_table extension work
      lots of clean-up, added d/s keys
      more work on GL_SGI_color_table, pixel transfer code clean-up
      applied Keith's fix for the Loki ice bug
      applied Bernd's changes to useBGR code
      version 8, added GL_EXT_texture_cube_map
      more clean-up
      added some missing NV token values
      added GL_NV_texgen_emboss
      wrapper for grTexLodBiasValue()
      minor updates
      cleaned up whitespace
      added a comment about FP overflows
      work on GL_SGI_color_table
      GL_SGI_color_table info
      dynamically allocate color table data, uses less memory
      more color table and imaging work
      added GLX_EXT_import_context, 1.3 misc clean-up
      minor clean-up
      minor clean-up
      updated RENDERER version
      removed old profiling code
      added _mesa_initialize_buffer/visual()
      call _mesa_initialize_context()
      more error checking.  Query GLX client vendor string
      removed extra /
      removed all DRI-isms
      Removed.  Use glxinfo instead.
      removed glxdpyinfo
      fixed tar file list of widget files
      fixed backslash typo in sunos configs
      implemented GL_EXT_histogram extension
      added GL_EXT_histogram
      removed software alpha flag from GLvisual struct
      made fxMesaCurrentCtx static
      conditionally include some fields in the context struct for DRI / non-DRI
      removed software alpha flag from GLvisual struct
      added a bunch of missing token values
      removed GL_HAS_GLEXT
      EnvMode GL_BLEND and tex invalidate changes from Daryll
      fixed the nested extern warning
      added check for tmuRam==4 for Voodoo2 (Bernd)
      toggle DD_LINE_SMOOTH in GL_LINE_SMOOTH case
      replaced buggy fxDDReadRGBAPixels() with read_R5G6B5_pixels()
      fixed glDrawBuffer(GL_NONE) bug
      more error reporting
      initial code for GL_EXT_histogram extension
      Err, initial code for GL_EXT_convolution, not histogram
      Added point UserSize, cleaned up point drawing code
      fixed bug in loop over screens
      a few Alpha tweaks (Michael Champigny)
      removed printf
      more GL_EXT_convolution work
      fixed PB bugs, removed MutablePixels, MonoPixels from GLcontext
      recover from offset=? condition
      recover from offset=? condition, code clean-up
      generate X86 dispatch code
      added glPointParameterf[v]SGIS()
      updated to version 10
      replaced components with internalFormat
      added Kevin's 32bpp Z buffer fix
      added tbuffer and compressed teximage extension functions
      bunch of updates to static_functions[]
      now generated from bin/glx86asm.py script
      changed ATTRIBLIST_SIZE to 32
      DeleteTexturesEXT was missing
      added glTbufferMask3DFX(), enabled texture compression functions
      Pass pixel format to xmesa_color_to_pixel().  Compute clearpixel without dither
      added const to compressed tex image functions
      added a bunch of imaging functions
      protect PFNGLCOLORSUBTABLEEXTPROC from multi-defined
      changed internalFormat to GLenum in TexImage4DSGI and all compressed textimage calls
      fixed alignment bug in _mesa_image_row_stride()
      minor tweaks for GLU and GLUT tar files
      moved a lot of Window-isms out of gl.h into other files
      include glext.h if GL_GLEXT_LEGACY not defined
      update for Glide3
      added some Window-isms formerly in gl.h
      defined APIENTRY if undefined
      added Window-isms previously in gl.h
      initial code for GL_ARB_texture_cube_map
      added GL_EXT_blend_func_separate
      replaced __CYGWIN32__ with __CYGWIN__
      replaced __CYGWIN32__ with __CYGWIN__, misc clean-up
      more work on GL_ARB_texture_cube_map
      more texture cube work, misc code clean-up
      initial work for GL_ARB_texture_compression
      changed glext.h include
      set default window pos
      register recent GL extensions for libGL compatibility
      enable and clean-up of paletted texture code
      replaced _mesa_base_tex_format() with base_colortab_format()
      more work on GL_ARB_texture_compression
      always define GL_ARB_multitexture (OpenGL 1.2.1 feature)
      changed allocation/initialization of API dispatch tables
      replaced  assertion with a conditional in _glapi_add_entrypoint()
      more work on GL_ARB_texture_cube_map
      Added GL_ARB_texture_cube_map
      applied Eero Pajarre's patch for Windows
      include glheader.h instead of system headers
      more work on GL_ARB_texture_cube_map
      initial check-in
      added cubemap demo
      initial work for GL_NV_blend_square extension
      added ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH() in _mesa_Fogfv()
      added VMS files to Makefiles
      added missing GL_BGRA case
      updated 3.2 section, added 3.2.1 section
      fixed bugs in fxSetupDepthTest()
      more work on GL_ARB_texture_compression
      updated OSF/1 CFLAGS (Jeremie Petit)
      added GL_EXT_packed_pixels extension, subset of GL 1.2
      clean-up in anticipation of glxext.h
      variety of updates to better conform to real GLX
      initial check-in
      added EXT version of vertex array functions
      added EXT version of vertex array functions
      changed dispatch of GL_EXT_vertex_array functions
      define GL_GLEXT_PROTOTYPES for glext.h
      GL_TEXTURE_BASE_LEVEL wasn't working correctly
      now using SGI's glext.h instead of original Mesa file
      use GL_CLAMP_TO_EDGE
      initial check-in
      added glxheads, manywin
      added dummy function to silence compiler warning
      moved #define of LLBL into assyntax.h file
      call glutInit() in main()
      removed non-windows exit() prototype
      added Motif info message
      revamped glCopyTexSubImage1/2/3D() to be more like glCopyTexImage1/2/3D()
      replaced fxTMReloadSubMipMapLevel() call w/ fxTMReloadMipMapLevel() to work-around subtex bug
      check for special cases of value = 0 or 1 in glAccum
      Added FREE(depth) in error clause before return
      applied joystick patch from Michael Champigny
      updated with version numbers, edits from Jon Leech
      added extension numbers for Mesa extensions
      fixed ColorMask test in update_rasterflags()
      updating mat ambient/emissive was broke. always compute ctx->EyeZDir
      disabled some ctx->NeedEyeCoords shortcuts
      init Point.UserSize
      fixed an assertion and minor bug
      fixed type checking error
      removed bad DrawDestMask assignments
      another patch for win32 joysticks
      draw a solid cube too
      print GL_RENDERER upon startup
      fixed GL_INTENSITY error checking bug
      initial check-in
      added texenv demo
      initial check-in
      added 3dfx demos
      initial check-in
      fixed compiler warnings
      initial check-in
      minor tweaks
      removed 3Dfx directory
      updated demos/ info
      added LoadRGBMipmaps2()
      added GL_EXT_texture_env_combine (Holger Waechtler)
      added aligned memory allocations (Gareth Hughes)
      initialize TextureStack matrices
      added code to initialize shininess tables
      minor updates
      added GL_EXT_texture_env_combine
      fixed two proxy texture typos
      added flat_multitextured_line()
      add separate specular color support in multitextured_rgba_points()
      don't add light's ambient color to BaseColor in gl_update_lighting()
      minor update
      several colortable fixes
      added GL_TABLE_TOO_LARGE to gl_error()
      fixed GL_ONE_MINUS_CONSTANT_ALPHA bug (geza)
      NeedEyeCoords if non-uniform scale transform
      enabled GL_POST_CONVOLUTION_COLOR_TABLE
      initialize some color table scale/bias factors
      added histogram sink test
      added GL_ARB_texture_cube_mapping and GL_EXT_texture_env_combine
      Holger's fixes for GL_EXT_texture_env_combine
      added texture cube map state save/restore
      changed MAX_WIDTH to 2048 to handle 2K textures
      reverted to old tessellator (GLU 1.1)
      reverted to old tessellator
      use indirect context, added error checking
      allocate 1-larger knot array in explode_knot() (Mesa bug 104881)
      first bits of convolution code
      replaced GL_CLIENT_ALL_ATTRIB_BITS with GL_ALL_CLIENT_ATTRIB_BITS
      test for MESA_DEBUG = silent in gl_error(), misc clean-up
      added extra check to optimized GL_RETURN case
      Fixed pixel color problem in general_flat_rgba_line(). Added debug code
      removed references to fx_regoff.h and fx_gen_regoff.c
      fixed AA sampling problem, misc clean-up
      another tweak to optimized GL_RETURN code
      changed version string to 3.2.1
      removed redundant configure file from tar list
      removed more redundant files from tar archive
      added: touch widgets-sgi/depend
      Fixed off-by-one error in compute_shine_table().  Minor code clean-up
      removed two bugs
      restore driver depth test enable/disable state in PopAttrib()
      initialize GenBit[STRQ] in init_texture_unit()
      a bunch of thread-safety changes
      call glutInit()
      added GLX multi-thread demo
      removed beta from version string
      added 3.2.1 info and final 3.3 info
      removed beta occurances
      lots of updates for 3.3
      imported from 3.2.1 release
      fixed glitches in zip file rules
      added placeholders for 3.4 and 3.5 info
      3.5 release notes
      added missing widgets-mesa/src/*.c files, updated version numbers to 3.5
      install glext.h too
      test against an epsilon magnitude in gl_rotation_matrix()
      glGetTexParameter(GL_TEXTURE_RESIDENT) didn't query device driver
      initialize texture priorities to 1.0, clean-up of _mesa_PrioritizeTextures()
      define GL_GLEXT_PROTOTYPES before including mesa_wgl.h
      removed GLU 1.2 symbols
      test handling of lots of textures
      updated by Alessandro Pisani
      added include/GL/mesa_wgl.h to tar file list
      fixed return val bug in glAreTexturesResident()
      updates from mesa3d at billbaxter.com
      updated Phil's email address
      added NULL parameter checking to API functions
      added GetInteger/Float/Double/Boolean/Pointer device driver functions
      bumped MESA_MINOR to 5
      updated version to 3.5
      fixed depth mask bug in fxddClear(DD_DEPTH_BIT)
      fixed segfault problem when using MESA_GLX_FX
      removed all dependencies on XMesa
      fixed some warnings (Alan H.)
      fixed typo which disabled simple_DrawPixels
      added missing radius multiplication factor to sphere code
      optimized glDrawPixels for glPixelZoom(1,-1) (Andrew Gee)
      implemented new ImageTransferState bitmask
      fixed image transfer errors
      basic glDrawPixels convolution implemented
      more convolution work, not done
      increased max convolution filter size
      added convolve.c to file list
      use ImageTransferState
      print UNDEFINED MODE over samples that are undefined in the GL spec
      changed quad vertex colors to be tessellation invariant
      finished separable filter functions
      moved convolution API functions to convolve.c
      enabling 1D convolution caused a problem
      added convolve.h header
      updated some comments
      don't free ctx inside _mesa_initialize_context() if error, caller does that.
      changed default Shininess to 6
      added MESA_FF_R8_G8_B8 (alpha filled to 0xff) format
      finished compressed texture support (Bill White)
      added more extensions testing code
      added _mesa_pack_float_rgba_span()
      more convolution work
      renamed some variables
      convolution for glTexImage[123]D now works
      added more resolutions (Andy Sloane)
      fbiRev and tmuRev weren't getting initialized when using Glide3
      added Glide3 bug fix info
      added new extensions
      initial check-in
      destBytesPerRow was uninitialized in make_texture_image()
      finished convolution
      enabled GL_EXT_convolution and GL_ARB_imaging
      added convolution, ARB_imaging items
      Fixed bug in fill_texture_image().  Clean-up in _mesaGetTexImage()
      added const to _mesa_select_tex_object()
      added tracing facility from Loki
      added tracing facility from Loki
      GetConvolution/SeparableFilter() now finished
      use clear_proxy_teximage() instead of MEMSET()
      passing proxy targets to glGetTexImage() should generate GL_INVALID_ENUM
      oops- forgot convolution for glGetTexImage.  Done now.
      replaced ## with CONCAT macro
      test for __bsdi__ for XFree86
      silence compiler warning (zoomY0)
      added device driver function for texture proxy testing
      newly organized Xlib driver files
      renamed Xlib driver files
      added xmesa_update_span_funcs()
      Removed ctx->Driver.LogicOp().
      added device driver changes
      added OSMesaCreateContextExt()
      Use OSMesaCreateContextExt() if using Mesa 3.5 or later
      added OSMesaCreateContextExt item
      changed gl_frame_buffer Xmax and Ymax to be exclusive boundary values
      removed ctx->Driver.Dither function
      added dither key option
      removed unused NEW_DRVSTATE flags
      added Dither, NEW_DRVSTATE info
      Removed immediate struct's maybe_transform_vb pointer.
      Win32 update (Chris Morley)
      Win32 updates (Chris Morley)
      added some casts (Takaaki Nomura)
      better fps computation
      terrain demo data
      better FPS calculation
      init mutex
      added format/type error check to draw_rgba_pixels()
      packing of GLushort packed types was broken
      replaced GL_RGBA with GL_BGRA for GL_UNSIGNED_SHORT_1_5_5_5_REV datatype
      added cursor key info to instructions
      explicitly init some texture vars
      updated comments
      use cpml.h on Alpha (AlanH)
      moved DECLARE_XFORM_GROUP macros outside of gl_init_3dnow_asm_transforms() to silence compiler warnings
      added more warning options to linux-elf-debug config
      added FPS calculation
      removed unused struct display_list_compilation
      added trace support (Loki)
      Added GL_MESA_trace extension (Loki)
      added GL_MESA_trace support
      added some missing const keywords in GetString
      set default window pos, changed 4 to GL_RGBA
      new comments, misc clean-up
      removed unused display_list struct
      added GL_ARB_texture_env_add
      use GL_NEAREST texture filter
      Replaced preprocessor symbol GL_MESA_TRACE with MESA_TRACE.
      First batch of OpenGL SI related changes:
      added imports.c
      renamed gl_copy_context to _mesa_copy_context
      include types.h
      new params for SpecificCompressedTexFormat() (Bill White)
      removed ctx->Texture.Enabled, use ctx->Texture.ReallyEnabled instead
      generalized support for N texure units
      added fps output
      support for N texture units
      textured line test
      more work to support N texture units
      improved to test alpha blending with texture palette
      drawing color-index image to RGB window was broken
      better handling of GLX_DEPTH_SIZE in glXChooseVisual()
      brighter colors
      update TriangleCaps in GL_POINT/POLYGON_SMOOTH cases
      if depthBits == 0, set DepthMask = 2^16 to fix per-fragment fog problems
      use gray background, memset TempImage to white to help with debugging
      added _mesa_get_teximages_from_driver()
      ported render begin/end calls from Mesa 3.4
      More fixes for MAX_TEXTURE_UNITS > 2.  Seems to work now.
      GL_MAX_TEXTURE_UNITS_ARB is now 3
      improved blending accuracy to fix Glean test failures
      call Driver.SetReadBuffer() in clear_color_buffers() to fix colormasked clear bug
      glPixelStore(GL_PACK/UNPACK_SKIP_IMAGES) wasn't handled correctly
      Rewrote get_static_proc_address().  It made mistakes in some situations
      use 1.1 glPolygonOffset instead of glPolygonOffsetEXT
      Changes for multitexture > 3, code clean-ups.
      polygon offset changes; added Visual->MRD field
      fixed a texture enable bug
      added a proper idle() function
      fixed 386/x86 mismatches, reordered a few configs
      generate separate libOSMesa library ala XFree86/DRI
      added libOSMesa.so info
      added idle() func, enable fog by default
      Put a white texel in each corner of each texture.
      signal NEW_TEXTURING if there's a Driver.BindTexture() function
      added incomplete() debug function/macro
      added a glFinish() call
      fixed value for GLU_NURBS_MODE
      clarified meaning of DEFAULT_SOFTWARE_DEPTH_BITS value
      removed X86/*.m4 files from tar archive
      added GL_BGR case to bytes_per_pixel()
      Enabled GL_EXT_secondary_color.  Fixed a bunch of typos in the dlist.c
      added info about GL_EXT_secondary_color and GL_EXT_fog_coord
      restored test for __alpha in _mesa_win_fog_coords_from_z()
      Basic work to support deep color channels:
      New colormac.h file for color-related macros.
      More GLchan color channel changes.
      more minor header file re-org (moved CONST, ASSERT, INLINE to config.h)
      renamed Current.ByteColor Current.Color
      set *t->flag = enabled in gl_extensions_add(), otherwise, extension booleans weren't set
      Removed UPDATE_IMAGE_TRANSFER_STATE, use _NEW_PIXEL flag instead.
      big overhaul of the demo, works better now
      removed debug printf
      added comment explaining why the demo is flawed
      Always call Driver.SetReadBuffer() in copy pixels functions, just to be safe
      print number of texture units
      update for separate libOSMesa.so lib
      added some missing backslashes
      GL_MESA_packed_depth_stencil extension spec
      added GL_MESA_packed_depth_stencil extension
      MESA_trace extension (Loki)
      update to GL_MESA_trace extension enums
      New implementation of antialiased lines.  Mesa should now pass the remaining
      hooked in new AA line code, minor clean-up
      added swrast/s_aaline.c
      removed obsolete VB vars
      added buffer test to xmesa_choose_point(), fixes X protocol error
      added a setup function for selection/feedback
      test ctx->RenderMode in chooser functions, removed dead code
      test ctx->RenderMode in chooser functions
      adjust x/y/zoffset parameters to fill_texture_image() by texture border width
      fixed mipmap setup bug
      updated with bug fix from 3.4 branch
      GL_BGR case was misplaced in _mesa_is_legal_format_and_type() (Jeff Hartmann)
      restored call to _mesa_select_tex_image() in _mesa_GetTexImage()
      patched to silence compiler warnings (Martin Lindhe)
      removed old TexImage() and TexSubImage() functions
      minor changes to silence compiler warnings
      renamed to histogram
      renamed imaging files to histogram since that's what's inside
      replaced s_imaging.h with s_histogram.h
      replaced _NEW_IMAGING with _NEW_PIXEL
      latest sources from Loki
      removed tr_attrib.c
      init WrapR to GL_REPEAT mode
      fixed another texImage mistake in _mesa_GetTexImage()
      bumped version to 3.5
      removed unused var
      Removed Driver.Color() and Driver.Index() functions.
      removed unused set_color() and set_index() functions
      more Driver.Color, Driver.Index updates
      rewrite of _mesa_win_fog_coords_from_z() so that both perspective and orthographic projection are handled correctly
      Replaced ctx->Point.Size with ctx->Point._Size
      added includes to silence warnings
      Minor header file changes to silence warnings.
      hacked demo to test multi-texture projective textures
      Mark Kilgard's reflection/shadow dino demo
      test texture palettes with multitexture
      added more programs
      Replaced Texture.CurrentD[] with separate Texture.Current1/2/3D vars.
      removed #include <tgmath.h>
      minor multi-tex changes, changed MAX_TEXTURE_UNITS to 4
      added some RENDER_START/RENDER_FINISH macros.  Minor GLchan changes
      minor clean-ups
      minor clean-ups (Klaus Niederkrueger)
      call RENDER_START/FINISH in read_color_image()
      call gl_update_state() if needed in glGetColorTable, glGetConvolutionFilter
      minor updates for widgets-mesa/
      fixed GLX_PBUFFER value
      map glIndex* functions to glIndexi() (glIndexub was looping back to glIndexub)
      Added MaxClipPlanes and MaxLights to gl_constants struct so T&L
      minor changes to fix compiler warnings
      Set ENABLE_LIGHT according to ctx->Light.Enabled whether or not
      added new source directories to LIB_FILES
      fixed a convolution pixel store bug
      Pass scale and bias values to _mesa_scale_and_bias_rgba().
      fixed parameter mis-ordering bug in Driver.StencilSpan calls
      changed error tests for depth and stencil buffers
      added -s option to control when glXSwapBuffers() is called
      call glFinish() before getting t1 time
      minor comments, clean-up
      Compute attenuated point size in a new pipeline stage.
      Initial work on GL_MESA_sprite_point extension.
      initial draft
      added LL_MESA_sprite_point extension
      fixed bug 123918: glXCopySubBuffer() y position off by one
      standardized GLX extensions header
      include glxext.h unless GLX_GLXEXT_LEGACY is defined
      more detailed texture error messages
      added GL_RESCALE_NORMAL cases
      added queries for convolution enables
      initialize convolution filters to zero
      define GL_ARB_imaging
      set default table size to 0, minor error check fix
      adjust texture size for convolution in glCopyTex[Sub]Image[123]D()
      _MultiTextureEnabled was being computed incorrectly
      generate GL_INVALID_OPERATION error for bad image format/type combinations
      minor changes to fix compile problems/warnings
      disabled GLX_SGIS_multisample (temporarily) since its tokens aren't defined
      clipping, fog, texture optimizations (Klaus Niederkrueger)
      disabled two optimized textured triangle functions because of texcoord interpolation errors
      improved precision of glReadPixels for colorbuffers < 24bpp
      tweaked truecolor dither kernel setup code
      added glReadPixels item
      use inverted Y coordinate for choosing dither value (conformance fixes)
      replaced GLfixed with GLfloat in _mesa_fog_ci_pixels() (Klaus Niederkrueger)
      fixed conformance problems in min/max and histogram result packing
      added some more GLX extension entrypoints (fix GLUT link problems when using glxext.h)
      updated glXSet3DfxModeMESA() prototype types
      Renamed texture object _P to _MaxLevel and _M to _MaxLambda.
      added more GLX extension function stubs
      minor fog clean-ups (Klaus Niederkrueger)
      Polygon._OffsetAny wasn't being computed
      set visual->MRD to 2.0
      minor GLchan-related changes
      include state.h
      Removed fixed.h (GLfixed now in mtypes.h, fixed-pt macros in mmath.h)
      missed a few color macro changes
      minor clean-up
      More color macro clean-ups.
      New point rasterization functions based on template code that
      added divide by zero checks
      added divide by zero check
      minor bug fixes
      changed some point functions, new picking code
      added divide by zero check
      added underflow check in validate_shine_table()
      Fixed bug in MinMagThresh computation.
      Added PointSize and PointParametersfv to dd interface, for completeness.
      initialize Driver.RenderPrimitive to _swsetup_RenderPrimNoop
      removed unneeded TXTR_COORD macros (Stephen Tse)
      fixed a small error in import_color(), include m_translate.h
      updated ctx->Driver functions for s/w setup
      test check-in
      fixed glTexSubImage error check bug #128775
      added dispatch override mechanism, used by trace extension
      Replaced struct gl_visual with struct __GLcontextModesRec from glcore.h.
      space-bar = pause animation
      added support for 4 texture units
      minor tweaks
      added DestroyHeads() to test clean-up code
      Lots of GLchan datatype changes.
      minor tweaks
      removed the unused/broken GL_PGI_misc_hints extension
      redo _mesa_PopAttrib() to call Mesa state functions so derived state is updated
      updated spec
      updates to the trace extension
      optimized lambda computation (Klaus Niederkrueger)
      fixed computation of _MultiTextureEnabled (again)
      only compute separate specular if texturing really enabled
      removed unused vars
      GL_SHADE_MODEL wasn't popped correctly
      fixed inverted Y coord for dithering in  write_span_DITHER_5R6G5B_ximage()
      last segment of AA stippled lines wasn't drawn
      fixed half pixel offset bug in aa point code
      added packed types to glGetMinMax and glGetHistogram error checks
      make gluPerspective() call glMultMatrix() instead of glFrustum() to get correct error semantics
      replaced frustrum with frustum
      added mesa_profile extern decl
      misc updates to match latest device driver changes
      added _mesa_adjust_image_for_convolution()
      Overhaul of texture image handling.
      added a number of debugging tests to be sure drivers handle texture images correctly
      updated texture image handling - STILL UNTESTED
      made some changes to the initialization of gl_texture_image fields
      new texture image routines work now
      remove unused functions
      check texture texel type before using optimized sampling functions
      added some code for proxy texture tests
      fixed divide by zero problem
      added divide by zero checks
      fixed texel indexing bug in _mesa_GetTexImage()
      GLchan fixes
      fixed bugs in _mesa_clear_depth_buffer(), bug #131366
      fixed bytesPerValue bug in OSMesaGetDepthBuffer()
      minor changes from XFree86 (dawes)
      minor updates to sync w/ XFree86 changes
      fixed RGB over/underflow bug for tiny triangles (bug 128969)
      added ctx->Driver.ResetLineStipple, misc clean-up
      added debug assertions to be sure drivers initialize all required function pointers
      fixed GL_SET bug
      gl_error clean-ups
      removed bogus assertions
      fixed bug in gl_validate_all_lighting_tables() found with CI mode conformance tests
      fixed a few CI mode span bugs and a dither bug
      fixed an assertion
      fixed a CI mode segfault, minor clean-ups
      removed a misplaced _mesa_logicop_ci_span() call
      updated with 3.4 and 3.4.1 entries
      fixed a CI-mode spot light conformance failure, but still not clear why
      minor scissor optimizations (Klaus Niederkrueger)
      added n-key to toggle nice fog, other minor clean-ups
      moved depth/index/stencil span packing into image.c
      Work in glGetTexImage() to return GL_COLOR_INDEX images.
      added MESA_GLX_FORCE_CI env var (useful for conformance testing)
      Changed FetchTexel() function pointer arguments.
      added a bunch of extensions (Ray Tice)
      Updated Driver.CopyTexImage[12]D and Driver.CopyTexSubImage[123]D functions
      plug in fallback teximage DD functions
      Implemented GL_SGIX_shadow and GL_SGIX_shadow_texture.
      demo of GL_SGIX_shadow
      added shadowtex
      added shadow extensions
      added extension #ifdef/#endifs
      more 3.5 updates
      fixed GL_LUMINANCE_ALPHA bug in fetch_2d_texel()
      removed some bogus code
      improvements to regions_overlap() function
      fixed bug in _mesa_stencil_and_ztest_pixels() found w/ DRI mga driver
      another fix in _mesa_stencil_and_ztest_pixels() for hardware fallbacks
      bilinear sampling of depth textures
      added a cube map error check
      tiny change to a gl_problem() call
      added filtering option, compute shadow tex size from window size
      fixed extension checking in _mesa_BlendEquation()
      Disable most extensions by default.  Drivers should enable them as needed.
      minor extension tweaks
      more tweaks
      colortable w/ size=0 wasn't handled correctly
      added fog coord clamping
      signal _NEW_TEXTURE in all teximage functions
      fixed glitches in error checking of imaging extensions
      GLvisual inside GLframebuffer is no longer a pointer, copy the struct instead.
      Now there's four possible display modes:
      removed some bogus error checks
      added s/w alpha buffer clearing
      fixed mistake in alpha buffer clear test
      fixed _mesa_getenv() weirdness
      changed AlphaFunc() ref from GLclampf to GLchan
      optimized lambda calculation (Klaus Niederkrueger)
      lots of gl_*() to _mesa_*() namespace clean-up
      more namespace clean-ups
      used indent to clean-up the code
      more clean-ups
      replaced gl_test_os_katmai*() with _mesa_test_os_katmai*()
      GL_EXT_fog_coord test
      GL_EXT_secondary_color test
      added missing ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH() macros
      fixed segfaults when tex unit 1 enabled, but not unit 0 (conform)
      added Driver.BlendColor() function, for completeness
      added some casts in the 16-bit GLchan macros, just to be safe
      _MinMagThresh[] should be GLfloat
      fixed a bunch of g++ warnings/errors.  Compiling with g++ can help find lots of potential problems
      added an assertion in the init code, just to be safe
      More g++ warning fixes.  Fixes for CHAN_BITS==16, it seems to work.
      test 16-bit/channel rendering
      changed bincoeff var to GLfloat, fixes bug introduced in previous check-in
      processed by indent to improve readability
      do fog interpolation if INTERP_FOG is defined, not when INTERP_Z is defined
      fixed RBGA ifdef typo
      use IROUND() instead of (int) cast (Klaus Niederkrueger)
      removed ARB_texture_env_add bool flag, use EXT_texture_env_add flag
      SGI SI GLU library
      added CCFLAGS and CPLUSPLUS definitions
      SGI SI GLU 1.3 header
      updated for SGI SI GLU
      added SI GLU info
      fix for gl_ztrick bug (Ove Kaaven)
      fixes to make old GLU code compile with new GLU 1.3 header
      depth images were broken by latest teximage changes, this fixes the problem
      sync'd with the latest changes in XFree86 (cygwin stuff)
      minor re-org in sample_depth_texture()
      renamed some vars, added a bunch of comments
      use new texture conversion code, still have to do image rescaling
      minor updates for 3.5
      added teximage rescaling code.  untested at this point
      clean-up in _swrast_choose_triangle()
      added GL_ARB_texture_env_combine, GL_ARB_texture_env_dot3
      minor comment update
      added GL_ARB_texture_env_dot3 enums
      enable GL_ARB_texture_env_combine and GL_ARB_texture_env_dot3 in _mesa_enable_sw_extensions()
      fixed Min/Max lod clamp bug
      minor tweak to visual setup for accum buffers
      added silly mesa_hack() function and -b option
      gears using GLX instead of GLUT
      added glxgears
      added GL_ARB_texture_border_clamp
      test texture wrap modes
      change glTexImage3D() internalFormat to GLenum to match other OpenGLs
      removed DO_GEOMETRY
      silence compiler warnings
      moved shadow code into teapot.c
      clean-up of tunnel demos
      set initial window pos
      double buffer by default.  if drawSmooth, use 50% alpha at top of polygon
      added rules for readtex.c and readtex.h
      texFormat was a random value in _mesa_store_teximage3d(), fix by undoing last changes
      fixed RGBA/RGB typo
      pass dst and src to CONVERT_TEXEL() macros to fix non-ANSI problems
      New mechanism for thread-safe GL API dispatch.  C-based dispatch is faster.
      remove TEST_L and JZ instructions; always jump through the _glapi_Dispatch pointer
      removed a debug message
      simplify test for specific texture formats
      use Image[i]->TexFormat to check for mipmap consistency
      check for startX off the right edge of the window
      adjustment to coverage sample positions
      compilation updates
      Removed DD_Z_NEVER.
      silence a compiler warning
      fixed compilation problems related to ctx->_Enabled removal
      texImage->FetchTexel wasn't getting initialized in _mesa_CopyTexImage1/2D()
      add texture border width to x/y/zoffset in gl[Copy]TexSubImage*D() functions to accomodate negative offsets correctly
      querying proxy texture image info was segfaulting because of NULL TexFormat pointer
      applied David's patch for parsing display's server:screen string
      clear pixel value wasn't byteswapped for scissored clears in clear_32bit_ximage()
      fixed wglUseFontBitmapsA() bugs (Frank Warmerdam)
      include fxdrv.h before other headers (Ivan Kalvatchev)
      another fix to clear_32bit_ximage()
      Updated bug report info in _mesa_problem()
      More texture image changes.
      fixed tmpTexelSize computation bug
      GL_INTENSITY case was incorrect in extract_float_rgba()
      replaced IntFormat GL_BGRA with GL_RGBA, fixes problems calling _mesa_base_texture_format()
      Added IROUND_POS() macro to mmath.h and use where appropriate. (Klaus Niederkrueger)
      fixed some divide by zero conformance problems
      Reimplemented glPopAttrib code for GL_TEXTURE_BIT. It now calls the
      Moved FRAC macro out of mmath.h into s_texture.c since it's only used there
      b-key toggles between textures w/ and w/out border
      added a work-around in _swrast_copy_texsubimage[123]d() to fix alpha channel problem in some DRI drivers (see comments)
      added missing 'break' in _mesa_TexEnvfv(), better error messages
      return hex string for unknown tokens in _mesa_lookup_enum_by_nr()
      fixed mis-ordered args to _math_trans_4f() in _tnl_fill_immediate_drawarrays()
      replaced ac->count with ac->count - ac->start in import_texcoord(), as in the other import functions
      another fix in import_texcoord()
      Fixed a number of minor GL_ARB_texture_env_combine/dot3 issues.
      minor optimization to dot3 computation
      added SecondaryColor functions
      added GL_ARB_texture_env_combine GL_ARB_texture_env_dot3 GL_ARB_texture_border_clamp (temporary until SGI updates this file)
      bug fixes, added missing state query cases
      removed glext.h hacks
      check-in of old updates
      Fixed a texture conversion problem: sometimes need to produce an intermediate
      improved a glBindTexture error message
      Minor fixes for Win32 (Karl Schultz).
      compilation fixes - texturing isn't working at this time
      fixed two mistakes in texture combine code
      assign texImage->FetchTexel only if NULL
      texture image code updates (not finished)
      fixed problems with texel component size queries in glGetTexLevelParameter()
      reordered a few blend enums
      added -i and -h options
      fixed typo in GL_ARB_transpose_matrix
      minor simplification in glGetTexLevelParameter
      added ShowAlphaBuffer feature - 'a' key
      added -Wundef to linux-debug compiler flags
      fixed a potential tex obj reference count problem involving multi-texture
      flush pipeline in WMesaSwapBuffers (Frank Warmerdam)
      fixed mem leak in XFreeFontInfo() call
      Added Makefile support for building libOSMesa16.so which is Mesa compiled
      added 16-bit/channel rendering info
      added _mesa_free_framebuffer_data()
      changed xmesa_visual and xmesa_buffer structs to directly include GLvisual and GLframebuffer structs instead of pointers
      added -display option
      sync with XFree86/DRI glcore.h, now identical
      applied Diego Santa Cruz's patch to triangulateRectCenter()
      minor XFree86 change
      changed parameters to _mesa_rescale_teximage2d()
      teximage updates.  seems to work fine.
      fixed vertex color problems
      minor clean-ups and warning fixes
      added assertion to catch infinite loop in _tnl_fixup_compiled_cassette() - might help Keith
      interpolate fog valus as floats, not fixed - fixed the swrast fog problem
      do FLUSH_VERTICES() in glGet*() functions to make occlusion testing work, perhaps needed for other cases, like GL_CURRENT_COLOR
      added info about driver status in 3.5
      sample plane equations at fragment centers, not lower-left corner
      set MRD to 1.0, fixes new Glean polygon offset failure
      fixed minor logic error in _mesa_StencilOp()
      make sure rotation plugged into sin/cos lookup tables is always positive, fixes FP exception
      don't call _mesa_win_fog_coords_from_z() if fog is disabled, fixes FP exception
      missed a GLfixed->GLfloat fog change
      check for tiny values in dir[] vector in calcposobjs(), prevents FP underflows later
      fixed Y flip problem in read_rgba_pixels() (Frank Warmerdam)
      fixed font/bitmap x-offset bug (Frank Warmerdam)
      fixed FP underflow problem in kfact computation
      minor clean-ups
      fixed loop bug in _mesa_ConvolutionFilter2D()
      check hardware type and use SETUP_SNAP if needed
      removed linux-386-opt-V2-glide config since FX_V2 symbol has no significance anymore
      minor clean-up
      fixed bad texture border test (sometimes caused a segfault)
      applied Jeff & Keith's optimization to write_span_mono_pixmap()
      RGBA mode GL_NAND was wrong, fixed 16-bit GLchan support
      fixed some divide by zero problems found w/ conform
      added check for zero-length spans, avoids potential problems later
      redo previous change in a more defensive way
      fixed scissor clear problem (Michael Saunders)
      New triangle rasterization code.  Store per-span initial/step values in the
      changed MAX_HEIGHT to 2048
      New IFLOOR, ICEIL macros from Josh Vanderhoof.  Fixes problems with
      moved some misplaced compiler flags from CC assigment to CFLAGS
      minor clean-ups and more comments
      increase FOG_MAX to 10.0 to fix CI conformace failure
      _mesa_pack_index_span() was broken for non-GLubyte datatypes
      Apply antialiasing coverage factor to alpha after texture application,
      added latest 3.5 changes
      generate INVALID_OPERATION if active texture != 0 in glMap[12][fd]
      use DEFARRAY, etc macros to work around 32k data limit on Macs (Tom Goon)
      removed (void) index; statement
      fixed Width/Height typo (Jeff Hartmann)
      fixed two GL error calls
      fixed a bad error string in ClientActiveTexture()
      added missing #undefs at end of file
      s/Height/height/ (Jeff Potter)
      initial support for GL_SGIS_generate_mipmap extension
      added multitexture support
      fixed a number of multi-texture line bugs
      a little more work on GL_SGIS_generate_mipmap
      SPARC assembly optimizations from David Miller.
      updated copyright
      added 3.4.2 changes, SPARC assembly
      import from 3.4 branch
      fix bug 426657: need to clamp RGBA before color table lookups
      added color table clamp bugfix
      prototyping some new stuff
      Finished GL_SGIS_generate_mipmap.  Should be trivial to enable in DRI drivers.
      added GL_SGIS_generate_mipmap
      Initial changes to allow Mesa and its fake GLX to be built into XFree86 libGL.
      change GLXContext and GLXFBConfig typedefs
      added some casts so that older glx.h can be used
      changed #if OLD_VMS to #ifdef OLD_VMS
      added missing GLX_SGIS_multisample tokens
      added void to __GLXextFuncPtr
      more work on XMesa / libGL integration
      removed unused var
      infrastructure for GL_ARB_multisample
      s/void */GLvoid */
      added new functions to GLX_functions[]
      removed some old DRI-isms
      added math/m_debug_clip.c
      more DRI libGL-related changes
      forgot to check-in yesterday
      added test for EXT_fog_coord
      added current raster fog coord and related code
      GLubyte -> GLchan fix
      fixed s/w alpha channel bug
      swapped in/out arguments to interp_func.  Improved comments
      replaced #ifdef 0 with #if 0
      applied Klaus Niederkrueger's CHAN_BITS==32 patch
      added minor comments
      removed some code in XMesaDestroyContext() which could lead to a segfault
      filled in a bunch of the function template strings
      minor clean-ups
      test for X86_FXSR_MAGIC to be sure we have the _fpstate.magic field, needed for SSE signal handler
      use unoptimized COPY_4UBV code on SPARC to avoid memory alignment problems (bug 430689)
      only copy GLubyte[4] as a GLuint on x86
      added COPY_4UBV bug fix
      add check for NULL malloc (from Stanford metacompilation analysis)
      fixed minor display list bugs (#431517)
      check for PB overflow in general_flat_rgba_line()
      Snap triangle x,y vertices to 1/16 subpixel positions.
      added latest line and triangle bug fixes
      disable pixel transfer ops for glGetTexImage(), per the spec
      don't compile glClientActiveTextureARB() into display lists, per the spec
      always feedback unit 0 texcoords
      removed old, redundant CurrentTransformUnit state var
      moved glTexEnvi calls to better logical locations
      fixed a +/- typo in the Y coord setup
      fixed an assertion
      Allow different max texture sizes for 1/2D, 3D and cube maps.
      disabled an assertion that doesn't seem to be needed
      fixed a comment
      define both GL_ALL_CLIENT_ATTRIB_BITS and GL_CLIENT_ALL_ATTRIB_BITS (spec and SI don't agree)
      replace 255 with CHAN_MAX
      Enable GL_ARB_texture_compression for XMesa/GLX driver.  Texture
      Added fi_type union typedef to glheader.h.
      Consolidated source files.  Since the re-org a number of source files
      replaced -g with -O3 under linux-x86
      tar file list updates for 3.5
      fix glDraw/CopyPixels w/ fog bug.  minor fog code clean-ups.
      more updates to tar file lists for 3.5
      add m4/ directory to tar file list
      added a few missing si-glu files
      removed beta from version string
      updates for the 3.5 release
      also print GL_MAX_TEXTURE_SIZE
      bumped version numbers
      added missing GLX_PBUFFER_WIDTH/HEIGHT tokens
      added GL_IBM_rasterpos_clip extension
      added initial 3.5.1 info
      optimized texture triangle updates (Klaus Niederkrueger)
      More raster fog coord fixes.
      support for RGB_565 pixels (Lars Andersson), untested
      fixed logic error in OSMesaMakeCurrent() for 565 pixel support (Lars Andersson)
      fixed segfault (NULL _ShineTable[0]) in shade_rastpos()
      added missing \'s
      fixed flat shading bug in affine_ and persp_textured_triangle() functions
      fixed two bad casts (Stephane Conversy)
      replace _NEW_PIXEL with _IMAGE_NEW_TRANSFER_STATE to catch color matrix changes
      silence a warning
      silence an uninitialized var warning
      replaced __inline with INLINE
      assorted changes for supporting GLfloat color channels (not done)
      added linux-osmesa32 config
      minor tweaks
      undo previous check-in (unfinished code)
      removed unused tiny triangle test code
      added latest 3.5.1 bug fixes
      more work on float colors (still not finished)
      assorted fixes for Win32 (Gerk Huisma)
      Various 16-bit GLchan fixes and Win32 fixes (Gerk Huisma)
      disable MMX blend code until it gets fixed
      include glu.h
      do some 16-bit GLchan computations in floating point (Gerk Huisma)
      Added Win32 memory debugging and fixed a few memory leaks (Gerk Huisma)
      GLchan changes
      added vms_x_fix.h to tar file list
      check for pixels=NULL in _mesa_store_teximage[123]d()
      GLubyte -> GLchan change
      don't clamp colors if GLchan==GLfloat in add_colors()
      fix some Win32 warnings (Gerk Huisma)
      rename ERROR macro to TE_ERROR (Gerk Huisma)
      don't make default teximage if glTexImage's pixels=NULL (Gerk Huisma)
      GLchan and GL_ADD clamping changes (Klaus Niederkrueger)
      Applied Klaus Niederkrueger's latest flat-shading clean-ups and some of my own.
      use GLAPIENTRY when declaring a couple function pointers (Gerk Huisma)
      fixes for proxy cube map texture images
      wglUseFontBitmapsA() update (Frank Warmerdam)
      replaced @ with ## or # to allow compilation on SunOS 5.6 (Bug 442742)
      added four missing GLX 1.3 tokens
      check if deleting _Current texobj in _mesa_DeleteTextures (Bart Barenbrug)
      undo previous check-in (restore the @ symbols)
      added latest 3.5.1 bug fixes
      more GLAPIENTRY fixes (Karl Schultz)
      fixed GLX context sharing bug (reported by Eric Plante)
      added GLX context share bug fix
      optimization to sample_lambda_2d() (Klaus Niederkrueger)
      Win32 fixes, etc. (Gerk Huisma)
      fixed NewState update bugs in _mesa_PopClientAttrib()
      removed some now bogus assertions (due to Klaus's latest optimization)
      fixed bug in DEST_4F template function
      fixed a glPush/PopAttrib bug related to texture binding (reported by Wes Bethel)
      added latest 3.5.1 bug fixes
      removed ss_interp.c
      rm .lo .la files for make clean
      fixed Solaris compilation problems (Bug 441817)
      added GL_EXT_texture_edge_clamp
      replace PI with pi
      Fixes for 32-bit GLchans: smooth/flat/textured triangles seem to work now.
      WindML driver (Stephane Raimbault)
      added WindML driver
      new debugging code
      better error reporting
      better error messages
      bumped version string to 3.5.1
      added textured line bug fix
      removed a semicolon in the USE() macro
      simple multi-window GLUT test prog
      OpenGL 1.3 support
      added OpenGL 1.3 tokens and prototypes
      minor updates regarding OpenGL 1.3
      version 3.5.1 updates
      added OSMESA_MAX_WIDTH/HEIGHT queries
      changes to allow the driver to compile - still doesn't work though
      fixed dangling pointer problem (Chris Burghart)
      added GLX_RGBA_TYPE, fixed GLX_LARGEST_PBUFFER typo
      prototype work for Pbuffer support
      update version to 3.5.1, minor OSMesa changes
      minor changes, prototype Pbuffer support
      fixed s/w alpha bug 458411
      WindML updates (Stephane Raimbault)
      add new WindML/Mesa man pages to tar file list
      hook compressed texture functions into driver
      minor re-org
      replaced some tabs with spaces
      minor optimizations for flat shading (Klaus Niederkrueger)
      more GL 1.3 and GLX 1.4 updates
      more GL 1.3 updates
      replaced gets() with fgets()
      silence minor warnings
      silence more warnings (Karl Schultz)
      more warning fixes (Karl Schultz)
      more warning fixes (Karl Schultz)
      Win32 updates (Karl Schultz)
      removed carriage return chars
      more warning fixes (Karl Schultz)
      compute dst->ooz, not dst->z in clip interpolation function
      fix fallback for stippled polygons, removed a debug message
      set GR_STWHINT_ST_DIFF_TMU1 flag in the init() function (fixes multiarb demo)
      GL_ADD texture mode was incorrect, renamed some vars
      added __glCoreCreateContext and __glCoreNopDispatch
      bump versions to 4.0
      updates for Mesa 4.0
      fixed some configure;make build problems.  cleaned-up copyright info.
      updated version info
      4.0 updates
      added missing tnl_dd files
      [rgba]shift values weren't always correct, only effected blended lines
      removed duplicate GLX_AUX_BUFFERS_BIT definition
      replaced x86_xform_raw4.S with x86_xform4.S
      updated extensions
      minor updates for new glext.h
      added GL_ARB_texture_mirrored_repeat
      updated for new glext.h
      added new entrypoints for 1.3
      removed src/mms_depend from tar file list
      minor updates
      added some missing Win32 files to tar file list
      warning fixes (Karl Schultz)
      added a few comments
      applied Andrew Lewycky's patches
      added latest bug fixes
      version bumps, etc for 4.1
      template for 4.1 release notes
      added a bunch of missing breaks for multisample gets
      added some (void *) casts to silence gcc 2.96 warnings
      init the hash table mutex in _mesa_NewHashTable (Won-Suk Chun)
      fix error in error checking conditional (bug 476846)
      added missing GL_MAX_TEXTURE_LOD_BIAS_EXT query
      casts to silence warnings from gcc 2.96
      define GL_VERSION_1_3 (doh!)
      added 1.3 function manglers
      added GLX 1.4 function manglers
      fix typedef problem found with MSDev 6.0
      BeOS thread support (Philippe Houdoin)
      improved AA subpixel sample positions (Ray Tice)
      added missing const to glXGetContextIDEXT() prototype
      new Python API generator scripts
      new, simpler API specification file
      API dispath updates
      included by glapi.c
      remove Glide and SVGA support from mesa-debug config
      remove -DMESA_TRACE from linux-debug config
      test for glXGetProcAddress()
      added test implementation of GL_ARB_window_pos
      added test implementation of GL_ARB_window_pos
      fixed a glitch
      added test implementation of GL_ARB_window_pos
      More span rendering optimizations from Klaus Niederkrueger
      minor update
      fixed bug 484142, missing backslash
      fix for bug 474479, remove Trace/* stuff
      fix some warnings
      don't use near and far for parameter names
      don't directly call glSecondaryColor*EXT() or glFogCoord*EXT()
      use glXGetProcAddressARB to avoid extension linkage problems
      use v2 color, not v0 in flat_ci_triangle()
      updated comment about CHAN_BITS
      dispatch changes to minimize hassle with XFree86 libGL
      prototype implementation of GL_ARB_depth_texture, GL_ARB_shadow, GL_ARB_shadow_ambient
      fixed potential array overwrite problem
      added enums for GL_ARB_depth_texture/shadow/shadow_ambient (temporary)
      call FLUSH_VERTICES() in _mesaTexParameterfv() (bug 491832)
      special case blend optmizations (Jeremy Fitzhardinge)
      fixed errors in feedback for glDraw/CopyPixels and glBitmap
      add dispatch offsets for vertex program functions
      vertex program check-in
      vertex program test program
      minor dispatach table changes
      planned extensions for 4.1
      4.1 extensions
      minor tweaks
      added GLX_SAMPLE_BUFFERS, GLX_SAMPLES
      oops, wrong value for GLX_SAMPLES
      s/glxGetProcAddress/glXGetProcAddressARB/
      added newline characters to trace strings
      free the NormalLengthPtr if non-null in _tnl_free_immediate()
      disable debug printf's, fix a crash
      vertex program fixes
      better api trace/log messages
      temporary hack for divide by W
      divide by W and clipping now work - in pretty good shape now
      rotating, lit torus via vertex program
      warp a quadmesh with a gravitational source via vertex program
      updated vertex buffer identifiers
      remove conditional in RESET_STIPPLE to fix conformance failure
      first checkpoint commit of Klaus's new span code (struct sw_span)
      first checkpoint commit of Klaus's new span code (struct sw_span)
      added some debug printfs (disabled)
      disable prototype matrix stack code
      Replace old matrix stacks with new code based on struct matrix_stack.
      bring 4.0.1 info into trunk
      remove some cruft
      have to clamp color+specular in _swrast_add_spec_terms_triangle()! (fixes Chimera bug)
      fix a comment typo
      added separate specular bug fix to 4.0.1 list
      some display list support for vertex program extension
      don't use macros, just expand them in-line
      regenerated from updated glsparcasm.py script
      fix bug 498555 - test for __GNUC__
      added info about which drivers work and don't work
      added some assertions to exercise glIsTexture()
      test multitexture and vertex arrays
      comments, clean-ups, added lots of const qualifiers
      comments, clean-ups, added lots of const qualifiers
      Vertex program checkpoint commit: converted all vertex attributes (color,
      infrastructure for vertex program attribute evaluators
      minor clean-ups
      Another vertex program checkpoint: clean-up of vertex attribute storage
      implemented vertex program point size control
      work-around for QNX compiler problem
      compile with -DNDEBUG
      texture env color wasn't used correctly in GL_BLEND mode
      added GL_BLEND/texenv color bug fix
      fixed GL_REPLACE / GL_RGB bug
      added GL_REPLACE / GL_RGB texture bug fix
      fixed int->float conversion bug in _mesa_TexEnviv()
      added glTexEnviv bug fix
      init whole immediate struct to zero
      Klaus's latest patches: change texcoord[3] to texcoord[4] everywhere
      Moved _mesa_VertexAttribPointerNV into varray.c.
      Changed gl_client_array's Enabled field to GLuint so it can actually
      set _NeedEyeCoords to zero if vertex program is enabled
      updated some comments
      added some comments
      don't apply scissor box in read_fast_rgba_pixels()
      added glReadPixels / scissor bug fix
      added vertex program sources
      oops, missed vpstate.c
      Removed api_compat stuff, there's a better way.
      remove api_compat.c
      removed api_compat stuff
      Fixed Windows compile problem with gl__unused413() function.
      added some missing error checks
      added latest bug fixes
      latest changes from Klaus
      added another test in regions_overlap() (Klaus)
      set PB->haveSpec in line functions that emit specular color
      changed PB (pixel buffer) to use texcoord[4] array
      new gl_mangle.h with korn shell script to regenerate itself from gl.h and glext.h (Ray Tice)
      updated gl_mangle.h (Ray Tice)
      Klaus's latest patches and some clean-up
      files to support doxygen (source code documentation)
      minor update
      added array_cache to TAGFILES
      Clean-up/renaming of the per-vertex attribute bits, specifically, the
      fixed array_cache.tag mistake
      added missing IsEnabled() cases for fog coord and secondary color arrays
      more info about building for Motif
      added fog and raster Z position controls
      LOTS of changes, building upon Klaus's work.
      More span clean-up, mostly texture-related.
      More span improvements.  Removed _mesa_write_monocolor_span().
      Still more texture/span simplification and clean-up.
      replace GLushort w/ GLubyte in DEST_4US template (fixes 16-bit GLchan problem
      return GL_RENDERER = Mesa Offscreen16/32 when CHAN_BITS = 16 or 32
      don't use affine/persp_textured_triangle funcs when CHAN_BITS > 8
      removed GLUT reference
      Clean-up and optimize alpha test code.
      sw_span can now hold x/y arrays of fragment positions - getting ready to
      Converted line drawing over to new span code, pb no longer used.
      removed references to removed source files
      Clean-up and remove dead code related to depth/stecil testing.
      fix evaluator data and matrix stack mem leaks (Robert Bergkvist)
      fixed typo in depth_test_pixels(), should fix reported VTK failures
      added an assertion, fix typo
      Daniel Borca's GLUT for DOS/DGJPP.
      Daniel Borca's new DOS/DJGPP driver.
      add new DOS files to tar file list
      silence compiler warnings
      now supports > 2 texture units
      minor clean-up
      updated GL_VENDOR string
      fixed float->int conversion. Fix conditional for 16-bit Z buffer path
      added SPAN_SPEC flag in smooth_multitextured_line()
      added alternate glBitmap code (disabled)
      uniformly pass texcoords as GLfloat [4]
      obsolete
      fixed a multitexture MatrixMode() bug
      added a glClear() error check
      finished up GL_ARB_window_pos
      GL_ARB_window_pos fog fix
      Finished up GL_ARB_depth_texture and GL_ARB_shadow.
      updated VERT_BIT_* definitions
      Finished up GL_ARB_depth_texture and GL_ARB_shadow
      added new depth_texture/shadow tokens (temporary)
      include string.h to silence warning
      use separate GC for SwapBuffers to avoid colormask problem
      updated to use GL_ARB_depth_texture, GL_ARB_shadow, GL_ARB_shadow_ambient
      removed dead code
      Optimized the interpolate_texcoords() function:
      Lots of improvements in the 2D texture sampling code.  Fewer function calls
      don't include s_fog.h
      Simplified fog code.
      undo some debug changes mistakenly checked in earlier
      Updated 1D/3D/cube mipmapping code to work like the new 2D code.
      check for initialized XMesaBuffer in XMesaGarbageCollect()
      Fixed out-of-bounds memory write problem (CONVERT_TEXEL_DWORD macro).
      DOS updates from Daniel Borca
      use official enum values for GL_ARB_depth_texture and GL_ARB_shadow
      updated comments for CHAN_BITS=16 or 32
      silence gcc warnings (Marc La France)
      syntax fixes for Solaris (David Dawes)
      minor patches from David Dawes
      new assertions, use INLINE macro (Klaus Niederkrueger)
      new DOS driver files from Daniel Borca
      DOS driver updates from Daniel Borca
      fix some minor warnings
      obsolete
      tweak file list for tarballs
      Removed the xm_context field from XMesaBuffer.  It was preventing N threads
      initial work for NV_vertex_program1_1 extension
      added a comment
      added _mesa_image_image_stride()
      fixed pointer arithmetic error in glCopyPixels
      fixed typo in CONVERT_TEXEL_DWORD for convert_abgr8888_to_ai88 textures (Michael Fitzpatrick)
      remove a debug printf
      updated comments. removed dead code.
      protect currentReadable reference with #ifndef GLX_BUILT_IN_XMESA
      Lots of changes related to framebuffer/window buffer resizing.  Basically,
      added XMesaResizeBuffers()
      buffer resize changes
      New mipmap lambda calculation.  Previously, trilinear filtering could
      added latest 4.1 bug fixes/updates
      rename ResizeBuffersMESA to ResizeBuffers
      fix Intel C compiler warnings (Gerk Huisma)
      Use MESA_PBUFFER_ALLOC/FREE macros to allocate all framebuffer and texture
      don't set plane mask for xm_buffer->gc, it causes an accum buffer conformance problem
      removed glSamplePass()
      latest extension headers from SGI
      removed temporary ARB_texture enum defs
      removed temporary GL_ARB_texture_mirrored_repeat enums
      replaced an assertion with a conditional
      use -O3 to compile osmesa16/32 targets
      DOS/Mesa driver updates (Daniel Borca)
      re-adding a lost file
      added 4.0.2 info
      bring over from the 4.0 branch
      Test implementation of proposed GL_EXT_shadow_funcs extension.  This just
      exercise GL_EXT_shadow_funcs
      more extensions
      removed a bad assertion
      fixed inequality in an assertion
      reset span.end to zero after rendering the span
      popping of user-defined clip planes was broken
      decrement Transform._AnyClip when clip planes are disabled
      Use invert_matrix_general() instead of invert_matrix_perspective() because
      use _glapi_get_context() instead of GET_CURRENT_CONTEXT(ctx); in buffer_size()
      Replaced ClipEnabled[] array and _AnyClip with ClipPlanesEnabled bitmask.
      DOS driver update
      minor optimization in GL_TRANSFORM_BIT popping
      GL_ARB_point_parameters support
      updates from Jose Fonseca
      faster blending (Jose Fonseca)
      DJGPP updates
      added GL_ARB_point_parameters
      minor re-org
      more blending updates from Jose
      return NULL from Fake_glXCreateContext() if dpy or visinfo are NULL
      added an error check
      call _mesa_ResizeBuffersMESA() in OSMesaMakeCurrent() to reconcile buffer size changes
      blend_add() modified for GLfloat channels
      fixup float/ushort conversion macros
      fixed 16/32-bit channel problems in CopyColor[Sub]Table()
      fixed 16 bit/channel problem in blend_modulate()
      use NULL lambda array for pixel texture
      Texture combine alpha subtraction was wrong.  Fixed 16-bit channel glitches
      added OSMesaGetProcAddress()
      fixed some 3D texture image stride bugs when using 16-bit channels
      latest 4.1 features and bug fixes
      repair Array.NewState flags (Michael Fitzpatrick)
      tranforming display list normals by non-uniform scale didn't work (Carlos Pereira)
      new MMX blend code (Jose Fonseca)
      Klaus Niederkrueger's latest changes: use INIT_SPAN() to init span primitive
      updated a few comments
      version 13 of glext.h - fixes GL_EXT_texture_env_dot3 token values
      fixes for corrected GL_DOT3_RGB[A]_EXT token values
      Jose's latest patch (GMBT_GEOMETRIC_CORRECTION)
      Disabled SPARC cliptest functions for now.  See bug report 544665.
      fixed problem with swap() function and GCC3 (patch 414404)
      restored writeAll = GL_FALSE test/assignment
      added some casts
      casts to fix GLint/GLuint mismatches
      added two (GLchan *) casts
      new casts
      C++ fix
      removed a stray debug printf
      pass context pointer to _tnl_free_immediate(), removed backref pointer
      Allocate a sw_span struct in the swrast context instead of allocating it
      fixed Width/Height cut&paste typo
      vertex program attribute array work
      added support for vertex program attribute arrays
      Vertex program attribute arrays seem to work now.  This includes fallbacks
      fixed assertion, replaced switch() with if/elseif
      check for vertex program mode
      added to-do list for the 4.1 release
      test program for NV_vertex_program + vertex arrays
      removed CVS $Log comments
      call _mesa_sizeof_packed_type() in _mesa_GetTexImage() (bug 547203)
      added GL_EDGE_FLAG_ARRAY_COUNT_EXT to glGetBooleanv() (Bug 548277)
      test mapsize against 1, not 0 in glPixelMap()
      added GL_TEXTURE_MAX_ANISOTROPY_EXT to glGetTexParameter (bug 548812)
      added GL_TEXTURE_MAX_ANISOTROPY_EXT to glGetTexParameteriv()
      replaced GLshort with GLushort in _mesa_sizeof_packed_type()
      Implemented GL_ARB_texture_env_crossbar.
      added linux-icc config (Marc.Dietrich)
      remove -xK flag from linux-icc
      Minor change to current raster position and texcoords.
      updated vertex program dispatch offsets to official values
      updated vertex program dispatch
      dispatch offsets for ARB_window_pos
      replaced experimental MESA_sprite_point with NV_point_sprite
      removed experiemental GL_MESA_sprite_point extension
      added GL_NV_point_sprite
      some initial work for fbconfigs/pbuffers
      added -axK to linux-icc CFLAGS
      added GL_NV_point_sprite entrypoints
      Added GL_NV_point_sprite functions: glPointParameteri[v]NV().
      added another spot-check to verify dispatch table correctness
      bring over Michel Daenzer's DRI changes
      use CONTENT() macro on line 205 (Josh Vanderhoof)
      more big-endian tweaks
      NASM fix for *_CONTENT() macros (Josh Vanderhoof)
      replaced 0xf and 0xf0 with TEXTURE0_ANY and TEXTURE1_ANY
      replaced hard-coded 4 with NUM_TEXTURE_TARGETS
      fixed point sprite bug
      s/GL_MAX_TEXTURE_RECTANGLE_SIZE_NV/GL_MAX_RECTANGLE_TEXTURE_SIZE_NV/
      version 15 of glext.h
      added missing texture cube map targets
      GGI driver updates (Filip Spacek)
      Added CPLUSPLUS for sunos5 entries
      Fix NASM problems, remove static var in SSE detection code.  (Josh Vanderhoof)
      Moved big/little endian code to glheader.h.
      New _mesa_debug() function to replace fprintf() calls.
      fixed a couple dumb mistakes
      more removal of fprintf() calls
      added Porting Information section
      Added ctx parameter to _mesa_debug()
      added missing ctx parameter to _mesa_debug() calls
      #ifdef-out unused big-endian texel fetch functions
      Implemented GL_NV_texture_rectangle extension.
      updated with 4.0.x -> 4.1 porting info
      updates to 4.1 features
      added GL_ARB_texture_env_crossbar
      test for GL_NV_texture_rectangle
      the GGI driver is up to date
      added some braces
      added a comment
      fixed window size initialization bug
      fix a comment
      Fix glXCopyContext bug
      Joe Krahn's context sharing test program
      Port glXCopyContext bug from 4.0 branch.
      new fix for initial window size problem
      Moved GL_PROGRAM_ERROR_POSITION_NV state (it's not per-program).
      put XSync in get_buffer_size(), plus clean-up
      Added new bitfields to tnl_eval_store for NV_vertex_program evaluators
      added support for NV_vertex_program evaluators
      set stage->inputs properly (I think), plus debug/comment updates
      checkpoint: updates for NV_vertex_program evaluators - a bit more left to do
      added some _mesa_debug() calls
      fix problems in _mesa_debug/printf()
      test program for NV_vertex_program and evaluators
      remove an unneeded statement in update_polygon(), added a comment
      finish-up vertex program evaluators
      removed extra right-paren
      pull in 4.0.3 docs to trunk
      fix CHAN_TYPE==GL_FLOAT compile problem
      Applied Matt Sealey's patch to remove/isolate all stdio.h function calls.
      prep for OpenGL 1.4 feature set
      indentation fix
      Implemented GL_EXT_multi_draw_arrays: glMultiDrawArraysEXT() and glMultiDrawElementsEXT().
      added unofficial dispatch offsets for GL_EXT_multi_draw_arrays functions
      added GL_EXT_multi_draw_arrays
      blending fixes for CHAN_TYPE==GL_FLOAT (Gerk Huisma)
      CHAN_TYPE==GL_FLOAT clamping.  call _tnl_MakeCurrent.  (Gerk Huisma)
      added Gerk Huisma's GLchan=GLfloat fixes
      Added divide by zero check (Jon Perry).  Remove dead vars.  Fix indentation.
      Overhaul of glRead/DrawBuffer() code.  Now, swrast->Driver.SetBuffer()
      added Read/DrawBuffer porting info
      re-enable X86 code
      applied Eric Anholt's patch for Athlon SSE
      Pedantic compiler fixes (Sven Panne)
      silence warnings
      fix patching offsets (Sven Panne) (patch 544637)
      fix warnings (bug 580528)
      define MAXSTRING and use vsnprintf to preven overflows
      misc updates
      struct sw_span is again allocated on the stack, but the arrays of span
      latest bug fixes
      re-check-in - something didn't work on the previous check-in
      changed normal stride from 12 to 16 bytes since normals are now GLfloat[4] (for vertex program)
      initialize SWvertex array to zeros
      change dest stride to 16 bytes instead of 12
      change GLvector3f to GLvector4f for normal transformation
      comments
      init tmp->Normal.size = 3
      fix up resulting FOGC and PSIZ when needed
      added (char *) casts
      added glutGetProcAddress() and GLUT_FPS env var option
      added glutGetProcAddress()
      bump lib version to 3.7.1, more make clean commands
      added new GLUT features
      added error checking for glCallLists() type parameter
      minor tweaks
      added multitexture and line stipple
      fixed array index typo in VERT_SAVE_RGBA() macro
      fix problems in FXgetImage()
      s/ProjectedClipPtr/NdcPtr/
      s/DrawDestMask/_DrawDestMask/
      replace Current.Color with Current.Attrib[VERT_ATTRIB_COLOR0]
      re-enable ASSIGN_NORM_GROUP( 3dnow )
      minor error message changes
      stub functions for glXAllocate/FreeMemoryNV()
      added glXAllocate/FreeMemoryNV()
      define GLAPIENTRY and GLAPI if not defined in gl.h
      fix up stuff accidentaly checked in a few days ago
      fixed compilation problem for swdd->SetBuffer
      s/5/NUM_TEXTURE_TARGETS/
      improved error string
      updates from Daniel Borca
      updated DOS file list
      updated wglGetProcAddress
      test for ARB_texture_cube_map when getting GL version(Ian Romanick)
      added GL_IBM_texture_mirrored_repeat (Ian Romanick)
      added OpenGL 1.4 entrypoints
      version 17
      removed glext.h fix-ups corrected in glext.h version 17
      GL_EXT_stencil_two_side extension, not 100% complete yet.
      added GL_EXT_stencil_two_side and GL_NV_fence
      hook-up dispatch for glActiveStencilFaceEXT()
      implemented glProgramParameters4[df]vNV()
      fixed glSampleCoverage typo
      added -l option to print interesting OpenGL limits
      minor updates to sync with XFree86
      don't query GL_MAX_CONVOLUTION_WIDTH/HEIGHT
      removed GL_MAX_CONVOLUTION_WIDTH/HEIGHT
      incorrect dimenstion passed to texture_error_check() in compressed teximage functions
      more detailed _mesa_error() strings
      optimizations to _math_matrix_rotate() (Rudolf Opalla)
      fixed pixel pack bug in glGetTexImage()
      doxygen comments
      more detailed _mesa_error() messages
      fixed _mesa_generate_mipmaps() target problem
      remove GLint t, pass target to _mesa_init_teximage_fields()
      pull-in changes from DRI/Mesa-4.0.4
      obsolete
      push and pop ESI (Daniel Borca)
      Updated BeOS support (Philippe Houdoin)
      two-side stencil info
      stencil two-side and 3dfx fix
      added src/BeOS/*.h to file list
      merge updates from 4.0.4
      updates from 4.0.4 (MESA_ycbcr_texture, APPLE_client_storage, etc)
      GL_MESA_pack_invert
      update from Daniel Borca
      merge gl_texture_image RowStride from DRI 4.0.4
      fix bug in index_mask() (bug 609837)
      new texture compression infrastructure
      _glapi_get_proc_address() now always returns a valid dispatch stub function
      better error checking related to YCbCr images
      bring over from 4.0.4 branch
      removed ctx->Polygon._OffsetAny - not really needed
      removed unused ctx->Polygon.OffsetMRD
      finished up GL_EXT_stencil_two_side
      record GL_INVALID_VALUE in glVertexAttrib*NV() if index > 15
      multiple GL_POINTS can now be rendered together into one fragment span
      Changed a number of context fields from GLchan to GLfloat (such as ClearColor).
      updated list of internal changes in 4.1
      restore ACCUM_BITS to 16
      GLfloat/GLchan updates
      more work on fbconfig/pbuffer support
      added XMesaCreatePBuffer()
      Pbuffers seem to work!
      pbuffer info and demo programs
      added pbinfo and pbdemo programs
      added GLX_SGIX_fbconfig and GLX_SGIX_pbuffer
      removed old comment
      GLX_NV_vertex_array_range and GLX_MESA_agp_offset extensions
      import from DRI CVS
      added missing dispatch code for new AGP-related functions
      s/glVertexAttrib4fNV/_glapi->Dispatch->VertexAttrib4fNV/
      change fog from GLfloat to GLfloat[4] (like other vertex attribs)
      fix a typo
      finally get rid of ctx->Texture._ReallyEnabled field
      disable XMesaCreatePBuffer() for XFree86
      XFree86 fixes (clib macro wrappers)
      disable wscx field in __GLimportsRec to be consistant with XFree86
      use new vertex bitfield tokens
      removed old VERT_* defines, use VERT_BIT_* instead
      removed old TEXTURE0_ defines
      disable _glapi_set_warning_func() call as it's not present in older libGLs
      check texImage->IsClientData before freeing memory
      updated porting info
      check mode for no change in _mesa_Read/DrawBuffer()
      undo previous check-in (check if ctx->Color.DrawBuffer == mode)
      Another round of glRead/DrawBuffer() clean-ups and simplifications.
      updated info about glRead/DrawBuffers()
      init swrast->CurrentBuffer
      check EXT_point_parameters before calling PointParameterfvEXT()
      added missing GLX_TRANSPARENT_RGB
      print transparent pixel info when using -v option (Gerd Sussner)
      Glide driver disabled by default
      context-related cleanups (ex: _mesa_notifySwapBuffers instead of _mesa_swapbuffers)
      disable wsPriv field in  __GLdrawablePrivateRec to match XFree86/DRI
      added latest 4.1 changes
      minor updates
      added missing glFinish()
      added gluTessNormal() dummy function
      allow querying of glut functions via glutGetProcAddress()
      remove 'beta' from GL_VERSION string
      make a few static function non-static for highpc.c
      update file list for tar file
      removed highpc.c and lowpc.c
      remove lowpc.c and highpc.c
      remove some crud
      surround vertex program code with #if FEATURE_NV_vertex_program/#endif
      applied Philippe Houdoin's patch, plus minor clean-ups
      patches from Gerk Huisma for float-channel rendering
      s/functionOffset/offset/
      new wrap modes (Ian Romanick)
      check for GL_EXT_paletted_texture
      only allow intformat==GL_COLOR_INDEX if EXT_paletted_texture is supported
      latest fixes
      fix bug in GL_MIRRORED_REPEAT_ARB (Ian Romanick)
      implement auto mipmap generation for compressed textures
      new _mesa_max_texture_levels() helper function - not used everywhere yet
      doxygen comments (Klaus Niederkrueger)
      GL_ATI_texture_mirror_once extension (Ian Romanick)
      added GL_ATI_texture_mirror_once
      add missing texture rectangle code in _mesa_DeleteTextures()
      fix refcount error in _mesa_DeleteTextures (Randy Fayan)
      remove debug printfs
      added glDeleteTextures refcount bug fix
      massive overhaul, see new keyboard options
      stex3d updated
      added missing code to copy texrect state
      Header file clean-up:
      include GL/glx.h
      remove all.h, mem.h, mem.c
      added skybox and options to load real images
      #include fixes
      wrapped a few more C library functions
      added some casts for mallocs (bug 630379)
      init a few more fields in _mesa_initialize_visual()
      set release date
      updated email addresses
      fix culling problem (bug 630649)
      update versioning/extensions for OpenGL 1.4 / Mesa 5.0
      preparation for 5.0 release
      bump OpenGL version to 1.4, Mesa version to 5.0
      minor updates
      more version bumps
      fix potential bug in _mesa_align_calloc/malloc (Frank van Heesch)
      test both XFree86LOADER and IN_MODULE to see if we're being built as the glx module
      s/BZERO/_mesa_bzero/
      remove debug printf
      s/sprintf/_mesa_sprintf/
      s/printf/_mesa_printf/
      don't call abort()
      remove some stray debug/abort code
      use wrappers from imports.h rather than stdio, stdlib, etc functions.
      s/getenv/_mesa_getenv/
      added a comment about future work
      fix a few HPUX problems
      simple GL_EXT_stencil_wrap test program
      fix the occasional missing triangle bug
      tweak pipeline flags for vertex programming
      list bug fixes in 5.0
      fix a typo
      renamed abs() function glu_abs()
      silence a bunch of warnings
      HPUX fixes
      fix for HPUX
      add support for GLX_ARB_multisample
      glxinfo update
      misc driver updates (Daniel Borca)
      FX driver fixes
      minor revision to 4.1 features
      support for GLX_ARB_multisample
      Updates for DOS driver from Daniel Borca.
      added GLUT and DOS updates
      fixed glTexParameteriv(GL_TEXTURE_BORDER_COLOR) error
      added some printfs to show current polygon mode and cull state
      minor multi-sample clean-ups
      a few dispatch pointers weren't set (glGenTexturesEXT for example)
      latest 5.0 bug fixes
      restored fixedEdges var (for VMS)
      surround fixedEdges decl by #ifndef NDEBUG / #endif
      added missing 1.4 glFogCoord functions
      added 1.4 functions
      added glFogCoord functions
      added glBlendFuncSeparate, glMultiDrawArrays, glMultiDrawElements
      added a few more missing 1.4 entrypoints
      implemented automatic code gen and individual function validation
      fixed a few category lines
      fixed active stencil face query bug
      minor clean-up
      fixed active stencil face query
      declare tempI as Int, not Real (Jon Perry)
      multisample / polygon smooth test
      added P4 / 3DNow bug fix
      fix one-bit error in float->int Z calculation
      fix float->int Z calculation in _mesa_span_default_z()
      added Z calculation fix
      changes to vishandle usage to prevent segfault in some circumstances
      compilation fixes
      removed -malign-jumps
      apply scale factor before clapping for DOT3
      latest 5.0 bug fixes
      html index to doc files, doxygen files
      omit dmesadxe.h from tar file
      minor update
      include imports.h
      init facing to zero
      last tweaks for 5.0 release
      optimized color buffer clear fallback (use mono-color span functions
      added comments about NULL mask parameter to span-write functions
      check for NULL span argument
      moved function declaration into the template, define the NAME to specify the function names
      template code for span functions for flat, memory-based frame buffers
      Lots of clean-ups.
      fix compilation problem (Philippe Houdoin)
      Overhaul of line drawing template code.  Make better use of sw_span mechanism.
      bump version to 5.1
      more clean-ups
      enable 1.4 extensions
      bring in a few AGP-related bits from 4.0.4
      put FEATURE_NV_vertex_program around vpstate.h include
      fixed cube texture auto-mipmap generation bug (641363)
      set default texcoords for glBitmap/glDrawPixels
      added missing ctx parameter to _tnl_flush_immediate() calls
      fixed bad clear value
      added missing call to _mesa_write_mono_alpha_span()
      use polygon offset to avoid multi-pass Z-fighting artifacts
      mask is always non-null in write_rgba_pixels and write_monorgba_pixels
      remove GLUTAPIENTRY from __glutCloseDownGameMode
      added const to casts
      declar vsnprintf for OS/2 and VMS (Evgeny Kotsuba)
      added GL_DEPTH_COMPONENT to two assertions
      updated to use ARB extensions
      fix bad _mesa_printf() calls (mesa bug 646753)
      clean-ups, comments
      properly handle very wide images
      added _mesa_print_texunit_state() for debugging
      let _mesa_problem() take variable arguments
      use/test GL_SGIS_generate_mipmaps if available
      bug fixes
      added RELNOTES-5.1 file
      improved error message
      comment changes
      DOS updates from Daniel Borca.
      DOS updates
      obsolete
      added new DOS files
      fix glCopyTexSubImage2D error checking (bug 661907)
      fix typo in rotation code (bug 659677)
      move t->Image[i]->Format test (bug 659012)
      fix some float/ubyte mistakes (Daniel Borca)
      added font-related symbols
      moved ADD_L instruction (Linus Torvalds)
      s/16/32/ (Josh Vanderhoof)
      remove ADD_L altogether
      fix fog stride bugs (Michel Danzer)
      OSMesa16 updates (Eric Plante), added _mesa_strtof()
      added _mesa_strtof()
      remove const from _mesa_strtof()
      initial code for render-to-texture
      prototype render-to-texture
      GL_NV_fragment_program
      First batch of code for GL_NV_fragment_program.
      fragment program execution
      simple GL_NV_fragment_program test
      fix glDrawPixels hang when clipping
      update CurrentStack in glActiveTexture if needed (bug 669080)
      another fix for glDrawPixels + clipping
      add some casts
      fix IBM CPP warning (Evgeny Kotsuba)
      use R/G/B/ACOMP indexes (Evgeny Kotsuba)
      use COPY_CHAN4 macro
      fix an assertion
      check for NV_fragment_program for glGetString(GL_PROGRAM_ERROR_STRING_NV)
      GL_ATI_texture_env_combine3 extension
      new X86 CPU detection code (Petr Sebor)
      GL_SGI_texture_color_table extension (Eric Plante)
      GL_SGI_texture_color_table
      added missing break (Leif Delgass)
      glGet(GL_CURRENT_SECONDARY_COLOR) should return 4 values, not 3
      improved solve_plane_chan() (Evgeny Kotsuba)
      push/pop color table state (Eric Plante)
      renamed dither-related preprocessor symbols (bug 536406)
      silence some warnings
      silence warning
      remove unused vars
      new texture enums (Leif Delgass)
      optimize compute_coveragef (Evgeny Kotsuba)
      remove bad assertion, fix minor bug in _mesa_lookup_rgba()
      Make GL_SGI_texture_color_table work per-texture unit.
      some casts (Evgeny Kotsuba)
      move ltor computation
      updated version test
      fix prefetch bug (Felix Kuhling)
      fix reported 3Dnow bug in _mesa_normal_tab (Felix Kuhling)
      update version numbers (Klaus Niederkrueger)
      print debug message when SSE is disabled by user (Klaus Neiderkrueger)
      updated with new functions (Bill Hoffman)
      use ARB flavor of window_pos extension
      unneeded (Daniel Borca)
      DOS updates (Daniel Borca)
      read/write files, not stdio (Daniel Borca)
      Use 'ndc' instead of 'proj' terminology.
      added a comment for _tnl_need_projected_coords()
      fix bug in popping texture color table state
      replace | with & (Laurent Desnogues)
      fix assertion in transfer_teximage() (Laurent Desnogues)
      disable normal xform code - see bug 673938
      replace _mesa_strtof() with _mesa_strtod()
      added cvsclean target
      added rules for making bz2 files and md5 checksums
      added OS/2 includes
      replace APIENTRY with GLUTAPIENTRY to be consistant with glut.h
      version bumps, etc
      more work on DEFINE/DECLARATION statements, symbol tables
      Implement parsing of texture instructions and prototype execution.
      fixed bad zoomed.zStep assignment (bug 687811)
      Replace some #defines with enums.  (Klaus Niederkrueger)
      latest DOS updates (Daniel Borca)
      updated DOS files for tarball
      fix zoom/clipping bug 689964
      use IROUND to convert depth values from float to int (bug 690728)
      added TEXTURE_*_INDEX tokens
      parser clean-ups
      Move span stuff from swrast.h to s_context.h (it's private).
      test if active texture is really changed in _mesa_ActiveTexture()
      silence warning with a cast
      added _mesa_find_line_column()
      parser clean-ups (use struct parse_state)
      use _mesa_find_line_column()
      casts to silence g++ warnings
      fix-up pixel_format enum stuff
      move matrix type enum out of GLmatrix struct
      parser clean-ups and simplifications
      pass the instruction to store_vector4().  implement KIL.
      assorted clean-ups and improvements
      set MAX_NV_FRAGMENT_PROGRAM_OUTPUTS to 3
      more parser test programs
      s/Parse_SwizzleSrcReg/Parse_VectorSrc/.  Parse +/- and absolute value syntax.
      More clean-ups.  Parse scalar/vector literals and defined identifier sources.
      fix float color interpolation (bug 694461)
      fix a cast (bug 694546)
      fix GLchan==GLfloat bug in solve_plane_chan (bug 694546)
      minor typo
      updated version
      document latest source code changes
      Killed mmath.[ch].  Moved low-level functions/assembly code into imports.[ch]
      remove some MESA_DEBUG stuff
      removed extra _mesa_memset16() prototype
      added extern "C" for C++
      added glFinish with comment
      better error messages (Leif Delgass)
      disable using optimized _mesa_sqrtf() for now
      added _mesa_inv_sqrtf() and INV_SQRTF() (Josh Vanderhoof)
      silence a GLchan=GLfloat warning with a cast
      fix GLchan=GLfloat problems
      added OS/2 version of IROUND (Evgeny Kotsuba)
      minor updates
      Documentation/website overhaul.  The website content and doc/ directory
      minor tweaks
      DOS update (Daniel Borca)
      test GLfloat OSMesa rendering
      fix/improve glAreTextures/ProgramsResident (Jose Fonseca)
      added Klaus Niederkruger, html clean-up
      DJGPP tweaks
      DJGPP tweaks
      include macros.h
      Clean-up of parser error handling/reporting.
      simple NV_fragment_program texture test
      added fptest1, fptexture, texrect, etc
      Fix up some fragment program texture enable issues.
      added -lstc++ info, convert more plain text to html
      clean-ups, re-org in preparation for fragment program work
      use TEX_UNIT_LOOP macro to coalesce some code
      Store partial derivative values in sw_span structure.
      move #include glheader.h before extern C namespace
      added comments
      Rewrite of fragment program named parameters, constants, etc.  Not done.
      remove/obfuscate all remaining email addresses
      fix GL_COLOR_MATERIAL bug
      subset spec (was subset.html in embedded-1-branch)
      add 2, not 4, in _mesa_get_compressed_formats()
      disable prototyped GLX_ARB_render_texture extension
      store correct EXP result
      replace _mesa_ prefix with _swrast_, remove s_histogram.[ch]
      histogram code move, remove s_histogram.[ch] from Makefiles
      remove some unfinished code accidently checked in
      fixed ColorMask problem for front-buffer rendering
      minor clean-ups, comments, etc
      fix EXP instruction and ARL-relative addressing
      DOS updates from Daniel Borca
      added entries for GL_ARB_vertex_buffer_object, GL_ARB_vertex_program, GL_NV_fragment_program
      GL_ARB_vertex_buffer_object, GL_ARB_vertex_program, GL_NV_fragment_program entrypoints
      proper dispatch for NV_fragment_program functions
      added _mesa_print_nv_vertex_instruction()
      disable normal xform code for now
      added GL_ARB_vertex_buffer_object functions
      minor updates
      fix texgen msg strings
      print GL_MULTISAMPLE_ARB
      Basic infrastructure for GL_ARB_vertex_buffer_object.
      remove duplicate line
      remove debug printf
      enable 1.4 extensions
      5.0.1 updates
      minor change
      added 5.0.1 info
      added MD5 checksums
      minor updates
      fixed bad texgen Q problem
      use gluSphere instead of glutSphere so we get texcoords
      reduce memory needed for vertex attributes (allocate on demand)
      minor clean-ups, const qualifiers
      clean-ups, const qualifiers
      Remove dead code.
      added gnumake/gmake info
      _mesa_realloc() fix
      comments
      New device driver hooks for texture object and texture image creation to
      use structure containment to derive xmesa_context from GLcontext
      s/gl_ctx/mesa/ and misc clean-ups
      fix default compare mode
      fix GL_LIST_MODE, GL_CURRENT_MATRIX_STACK_DEPTH_NV and GL_CURRENT_MATRIX_NV queries
      GL_FOG_COORDINATE_SOURCE_EXT query was missing
      GL_SECONDARY_COLOR_ARRAY_SIZE_EXT query was returning stride
      improve some error messages
      minor initialization fixes
      added GL_FRAGMENT_PROGRAM_NV (Alan Matsuoka)
      clean-up current state initialization
      added GL_PROXY_TEXTURE_RECTANGLE_NV to is_proxy_target()
      fix double-memory free in XMesaDestroyContext
      simplify CEILF, FLOORF, FABSF stuff
      some texture compression odds & ends
      more memory-related fixes
      Rework gl[Copy]Tex[Sub]Image() error checking so that all level, width, height
      fragment program named constants and named program parameters basically work now
      test glProgramNamedParameter4fNV()
      Added gcc-sl target and shared lib shell script (Philip Brown)
      fix color array mixup
      fix typo
      minor parser fixes, program print-out fixes
      Demo of per-pixel lighting with a fragment program.
      initialize attribute vectors to 4 components
      scalar source reg can be a scalar or vector literal
      added _mesa_realloc()
      Added ctx->Texture._EnabledCoordUnits bitfield.
      fix some problems parsing scalar source arguments
      add missing return in raster_pos4f()
      fix string compare bug
      basic work on GL_ARB_vertex_buffer_object
      Added GL_ARB_vertex_buffer_object tokens
      Added function pointer typedefs for GL_NV_fragment_program
      add support for swap and sync extensions (Ian Romanick)
      fix typo
      fix typo
      remove a const
      Some groundwork for GL_ARB_vertex/fragment_program.
      fix depth texture tex env bug (#719903)
      checkpoint: more infrastructure for GL_ARB_vertex/fragment_program.
      temporary GL_ARB_fragment_program tokens
      Added arbprogram.c to Makefiles.
      added LoadProgramNV and RequestResidentProgramsNV
      Demo of using GL_ARB_vertex/fragment_program to do per-pixel lighting.
      clean up usage of extension function pointers
      fix some glBindProgramNV/ARB details
      more GL_ARB_vertex/fragment_program bits
      more GL_ARB_fragment_program tokens
      minor re-org of program matrix, program local parameter limits
      fix .c / .h typo
      use INDEX8_SPAN_ARGS
      fix minor warnings
      casts
      silence warnings
      added a cast
      silence warning
      alias ARB/NV program functions where possible
      added GL_EXT_depth_bounds_test
      Implemented GL_EXT_depth_bounds_test.
      fix DDX, DDY instructions (probably not finished though)
      remove unneeded unsigned int / zero comparisons
      remove unneeded fragment program test
      fix function prototypes for manging (Bill Hoffman)
      updated with newest functions
      fix glIsBufferARB() prototype
      new comments (Klaus Niederkruger)
      LOG2() clean-up (Klaus Niederkrueger)
      don't try to clear non-existant buffers
      updated for glIsBufferARB
      change global_grid_u0/u1 to REAL (Eric Cazeaux)
      Ian's buffer swap extensions
      Use ctx->Const.MaxTextureImageUnits and MaxTextureCoordUnits in more places.
      more docs (Klaus Niederkruger)
      updates from Ian
      minor updates
      fixed delete statement (bug 721765)
      added comment about MAX_WIDTH and textures
      clamp fog color
      fix MapBufferARB() return type
      fix scalar literal parsing glitches
      initialize condition codes before running program
      added GL_ARB_vertex_program functions
      remove temporary stuff
      version 18 of glext.h
      updated with new glext.h stuff
      added a couple casts
      autoconf/automake updated for newer versions (bug 734362)
      a couple casts (Evgeny Kotsuba)
      check buffer->UseSoftware*Buffer in _mesa_free_framebuffer_data()
      Version defines.  include "version.h" in a few files.
      glBindProgramARB dispatches to glBindProgramNV (remove _mesa_BindProgramARB).
      added MESA_VERSION() macro, etc (Holger Waechtler)
      silence some warnings (Evgeny Kotsuba)
      s/_T_VBXFORM_H/_T_IMM_EXEC_H/
      applied Ian's misc patches
      display both YCbCr and RGB image (Ian)
      include readtex.h
      misc changes from testing
      minor clean-ups
      added GL_NV_light_max_exponent
      minor error check tweaks for NV_light_max_exponent
      setup default programs for GL_ARB_vertex/fragment_program
      Replace most mklib.* scripts with a new uber-mklib script with more features.
      Remove a bunch of really old/obsolete configs.
      latest updates for 5.1
      updated release steps
      remove test file
      Added GL_LIB_NAME / GLU_LIB_NAME so the default name for the libraries
      updated rule for make OSMesa lib (Brad King)
      s/DrawBuffer/ReadBuffer/ in _mesa_ResizeBuffersMESA (PHil Brown)
      Added GL_EXT_texture_rectangle (identical to GL_NV_texture_rectangle)
      comments regarding window sizing
      updated old-style makefiles for new tree
      fixed secondary color typo
      add missing vertex attribs in _tnl_copy_to_current()
      update for new directory tree
      fix GL_CURRENT_VERTEX_ATTRIB_ARB queries
      GL_EXT_texture_rectangle, directory tree info
      fixes for x86 builds
      Further makefile clean-up, concentrated on the x86 assembly files.
      moved from main/ to glapi/
      clean up _tnl_copy_to_current()
      added missing types to _mesa_VertexAttribPointerARB()
      new buffer/context visual compatibility test (Phil Brown)
      added null pointer check in tnl_copy_to_current()
      s/ATTRIB_h/ATTRIB_H/
      added new GL_ARB_v_p instructions
      More clean-ups and re-org
      Implemented GL_ARB_occlusion_query (not 100% finalized).
      GL_ARB_occlusion_query demo.
      fix typo, added space/anim control
      GLUT for OS/2 (Evgeny Kotsuba)
      fix include paths (Daniel Borca)
      DOS updates for new tree (Daniel Borca)
      add a few more digits to PI to make Evgeny happy
      DOS updates (Daniel Borca)
      Simplify extension string handling.
      tweak checks for display list zero
      Implemented GL_ARB_texture_non_power_of_two (except for auto mipmap generation).
      added GL_ARB_texture_non_power_of_two
      GL_ARB_texture_non_power_of_two test program
      bigger window
      fix int/float conversion for GL_TEXTURE_PRIORITY in gl[Get]TexParameteri[v]
      minor fixes in _mesa_TexGen[id]v() to silence valgrind
      fix conformance glitch introduced with NPOT extension
      added GL_VERSION check
      DOS updates from Daniel Borca
      more DOS updates from Daniel Borca
      Fix up some loose ends from the last big check-in.
      Fix up FEATURE_* defines in config.h
      Karl Rasche's ARB_vertex_program parser
      Numbered productions for ARB_vertex_program grammar (Karl Rasche)
      Initial implementation of GL_MESA_program_debug - a vertex/fragment program
      test program for GL_MESA_program_debug extension
      vertex/fragment debugger extension
      tokens, functions for GL_MESA_program_debug
      Implement debugger callback, etc for vertex programs.  Misc clean-ups.
      Restore more code lost during last big merge.
      New interactive debugging code - vertex or fragment program.
      remove printf
      fill in chapter 5, 6 text
      fix GL_SGI_texture_colortable bugs
      fix a few more glitches from last big check-in
      if texture color table is enabled, use the color table's format to evaluate the texture env function
      don't include malloc.h (bug 776574)
      replace assertion with conditional
      Updates from Daniel Borca.
      Added TOP to point to top of tree to find includes and libs
      updated item 1.4
      added info about Mesa-newtree
      placeholder files
      remove temporary prototypes
      fix a bunch of warnings
      prep for OpenGL 1.5
      added arbfrag/vertparse
      fix some broken links
      Re-org of register files for vertex/fragment programs.  Will be easier to
      register file changes
      minor code improvement in glGenTextures
      simplify some python code
      glPixelMap[f/ui/us]v()'s size parameter is GLsizei, not GLint.
      Fix glPixelMap: s/GLint/GLsizei/
      glPixelMap fix
      DOS and glide driver updates from Daniel Borca
      updates from Daniel Borca
      restore gcc-sl config
      patch to import Jon Smirl's work from Bitkeeper
      glDrawPixels(GL_DEPTH_COMPONENT) with glPixelZoom didn't work
      test program to exercise glDraw/ReadPixels(GL_DEPTH_COMPONENT) and pixel zooming
      indentation fix
      regenerated for GL_IBM_multimode_draw_arrays
      fix PFNGLMULTIMODEDRAWARRAYSIBMPROC
      latest extensions and updates
      make _glapi_get_proc_address() return NULL if function doesn't start with gl prefix
      Added OpenGL 1.4's per-texture LOD bias.
      space bar to toggle animation
      remove a bunch of the extension definitions from the legacy section
      added GL_EXT_texture_lod_bias to OpenGL 1.4 feature list
      remove unused bytes variable
      Updates from Daniel Borca
      Move clamping of texture LOD bias to texture application time.
      fix minor warnings with casts
      s/GLuint/GLint/
      Moved some shared vertex/fragment program code into new program.c file.
      added minor assertion
      added program.c plus minor fixes
      update parameter indexing
      always assign texcoord[i][3] to silence valgrind
      s/DP4/DP3/
      s/DP4/DP3/
      fix fog.End error
      update from Daniel Borca
      add missing glNormal, glVertex calls (Michal Wozniak)
      fix-up CXXFLAGS (bug 757292)
      pbuffer info
      added GLUT_CFLAGS to compile GLUT with -fexceptions (bug 726223)
      remove -malign-loops=2 -malign-functions=2 (bug 738020)
      add -lCrun for sunos5, added sunos5-smp (Greg M)
      include math.h
      ARB_point_sprite is not part of OpenGL 1.5
      fix a few typos for linux-solo
      added fbdev/DRI and MiniGLX info
      assorted updates
      a little more MiniGLX info
      Added new opcodes for ARB_fragment_program, like ABS, CMP, TXB, etc.
      import 5.0.2 info
      move SWIZZLE_ZERO/ONE tokens to program.h
      implement SWZ and TXB.  some code clean-up
      files for CVS to ignore
      Updated for 5.0.2 release.
      updated info for Mesa 5.1
      update directories
      misc updates
      move GL_MESA_program_debug code into program.c
      fix typo
      more debugger development
      use GL_GLEXT_PROTOTYPES
      update sunos5 targets (Greg Menke)
      SunOS tweaks (Greg Menke)
      Remove all the swap_control stuff from glxgears demo (restored to vers 1.2).
      fix timer bug (801485)
      remove last of mklib.solaris stuff
      Added prototypes for internal functions to bufferobj.h.
      remove duplicate sunos5-smp entries
      plug in vertex buffer object functions
      display list support for ARB_vertex/fragment_program
      Allow glTexImage1/2/3D to specify width/height/depth = 0.
      allow glTexImage1/2/3D width/height/depth = 0
      Some work on ARB_vertex_buffer_object.
      minor simplification
      fix up FEATURE_* defines
      set ARB_vertex_buffer_object in _mesa_enable_sw_extensions()
      move some functions, add some comments
      move a comment
      changed a comment
      More work on ARB_vertex_buffer_object.
      added call to _mesa_enable_1_5_extensions()
      new casts
      GL_ARB_vertex_buffer_object working now, at least for non-indexed
      test GL_ARB_vertex_buffer_object
      Implement GL_ELEMENT_ARRAY_BUFFER_ARB for buffer objects.
      Exercise the GL_ELEMENT_ARRAY_BUFFER_ARB path
      s/GL_READ_WRITE_ARB/GL_WRITE_ONLY_ARB/
      unbind vertex arrays from buffer in DeleteBuffersARB()
      Fixed/added some error checks.
      remove unused Target field from gl_buffer_object
      updated with new extension info
      assorted updates
      Remove FLT_MIN definition to imports.h
      removed unused NoRaster and Rendering fields from GLcontext
      version tweaks
      remove MESA_TRACE stuff
      Move context ASSERT/FLUSH macros to context.h
      Move macros from header to C file.  Updated some comments
      s/_backface_sign/_BackfaceSign/
      tweak an error message
      Move away from using the ctx->_TriangleCaps bitfield.
      a few less occurances of _TriangleCaps
      added note about _TriangleCaps
      minor updates
      summary of Mesa's extension enum values
      Assorted casts to silence g++ warnings.
      casts
      apply patch #740145 from SourceForge
      remove a _TriangleCaps reference
      minor clean-ups
      new enum values for GL_MESA_program_debug
      enum updates
      added info about vertex/fragment program debugger
      GL_ARB_point_sprite demo
      updates from Daniel Borca
      improved 'clean' targets (Otto Solares)
      fix error message (bug 816876)
      Eric Plante's updates for OSMesa16.
      old/unused
      remove bogus _swrast_texture_table_lookup() call (Eric Plante)
      add switch defaults to silence warnings
      don't use color table's format to determine texture env functions
      don't request accum buffer in color index mode
      s/GL_COLOR_INDEX/OSMESA_COLOR_INDEX/
      fix some sphere texgen bugs (reported by Daniel Borca)
      added GL_ARB_point_sprite demo, texgen bug fix
      minor fix in buffer_object_get_target()
      bufObj->Access wasn't being set
      exercise more vertex buffer object functions
      remove some cruft
      minor tweaks
      initialize GetBufferSubData pointer
      fix minor warnings from g++ 3.2
      protect whole file with #ifdef FX
      added GL_ARB_point_sprite tokens
      obsolete
      update list of files for tarballs
      add .dsp files to tarballs
      fix CVS info
      Move initialization of default ctx->Driver.buffer-object-functions into
      fix an error message
      wrap in #ifdef FX / #endif
      added more debug code and tests
      Initial work for bounds checking of vertex arrays and vertex buffer objects.
      s/GLuint */GLvoid */
      added MorphOS instructions (Matt Sealey)
      Pass $(MKLIB_OPTIONS) to mklib script to allow setting extra options.
      minor fix to glutSetColor prototype (Evgeny Kotsuba)
      prefix MAKELIB with TOP in default case (Matt Sealey)
      another tweak from Matt Sealey
      fix AccessViolation bug (#835861)
      added wgl bug fix
      s/xoffset/zoffset/ (Cedric Gautier)
      added glTexSubImage3D bug
      fix comments
      s/GGIMesaContext/ggi_mesa_context_t/
      Redo array element checking for vertex array buffers.
      Ville Syrjala's logic op patch
      updated 5.0.2 tarballs
      added Ville's RGBA blend fix
      no-op glAccum in selection/feedback mode (per Mark Kilgard's updates to the 1.5 spec)
      added glAccum bug fix
      remove GL_GLEXT_LEGACY
      fix texgen bug 597589
      fix comments about Cr==V, Cb==U
      Some groundwork for supporting GLhalf datatype.
      added missing <type> error checking in read_index_pixels()
      added reference to original half/float conversion code
      fix a bunch of warnings
      reformat comments for 80-column viewing
      s/byte/GLubyte/ and other minor clean-ups
      fix some bugs in computation of DDX, DDY commands
      GL_ARB_point_sprite is not part of 1.5
      pass ctx to fetch_vector4_deriv()
      added OpenGL 1.5 tokens and functions not yet put into glext.h
      include malloc.h to get ptrdiff_t for vertex buffer object's GLintptr
      added GL_ARB_vertex/fragment_program and new vertex code
      minor error string improvement
      whitespace and comment changes
      fix YCBCR assertion and disable assertion in auto-mipmap code
      set the VB->FogCoordPtr with the other legacy pointers
      updated list of new/old src files
      generate the OpenGL 1.5 entrypoints
      minor cast fixes
      oops, fix glitches in 1.5 functions
      updated after APIspec fixes
      Remove unnecessary usage of __FUNCTION__.
      added a cast
      remove __FUNCTION__ hack
      silence a warning
      fix a lot of warnings found with g++
      casts for g++
      fix assorted g++ warnings
      Otto Solares' patch to add radeon 9200
      added SoftGL link
      current raster color index should be GLfloat
      updated CVS info
      added link to Chromium
      updated NetReality -> Pansophica
      fix an error test
      if id==0 in glGetQueryObject, raise GL_INVALID_OPERATION
      more tweaks to error checking and reporting
      Fix generic/conventional vertex array glitches.
      Use XPutImage() for glDrawPixels() and XCopyArea() for glCopyPixels() when
      s/texUnit->Enabled/enableBits/ (Karl Rasche)
      updated list of files since merging vtx branch
      fix __sparc_v9__ test to exclude linux (bug 852204)
      added description of faster glDraw/CopyPixels code
      Fix some problems with glDrawElements and vertex buffer objects.
      Port over changes from XFree86/Mesa 5.0.2, mostly to silence compiler warnings.
      don't use new glDrawPixel routines in Xserver
      press f key to toggle front/back drawing/copying
      Fix a front/back CopyPixels glitch.
      don't include GL/gl.h, define GLAPIENTRYP if not defined
      include glheader.h first, to prevent warnings about GLAPIENTRYP being redefined
      s/APIENTRY/GLAPIENTRY/ for 1.5 functions
      basic API work for FBConfigs & Pbuffers
      glFBDev driver from embedded-2 branch.
      Basic info about the glFBDev interface/driver.
      tweaks for SunOS
      conditionally define GLAPIENTRYP to prevent some warnings in linux-solo
      added drivers/fbdev/glfbdev.c
      Update NEED_SECONDARY_COLOR macro to test if either vertex/fragment
      added fbdev/glfbdev.c to tarball list
      replace gl/mesa_wgl.h with GL/mesa_wgl.h (Simon Goodall)
      check for null dpy in Fake_glXGetConfig()
      fix visinfo typo
      new Makefiles
      lists of common source files
      improved makefiles
      makefile improvements
      new makefiles
      uncomment-out tdfx driver
      updated list of files
      applied Felix's patch for configuration system
      remove TexCoordInterleaveFactor.  Clean up _mesa_InterleavedArrays()
      check swrast->NewState in Draw/CopyPixels functions
      Updates from Philippe Houdoin.
      added BeOS files, src/mesa/sources file
      init vars to silence warnings
      call _ae_destroy_context() and _ae_invalidate_state() from the proper places
      fix some matypes.h glitches
      fixed bad touch command
      Assorted mklib tweaks.
      fix GL_MIN/MAX w/ glBlendFuncSeparate bug (Ian Romanick)
      s/Makefile.X11/Makefile.solo/ for make clean
      updated location of config.h
      updates for 5.1 release
      added MD5 checksums for 5.1 tarballs
      added link to RELNOTES-5.1
      Version bumps, etc for Mesa 6.0
      the optimized UNCLAMPED_FLOAT_TO_UBYTE macro was broken
      updated CVS info
      added info about S3TC texture compression
      added -fPIC -fomit-frame-pointer etc to sunos5-gcc configs (bug 868737)
      latest updates
      whitespace changes and new comments
      added missing descrip.mms files to tar file list
      remove BMRT link
      added Aqsis link
      updates for 6.0 release
      reindent the code
      set StringPos field in parse_vp_instruction (Mesa bug 873011)
      Change < to <= when testing counts against ctx->Const.MaxArrayLockSize,
      re-enable SSE code, except for the _mesa_sse_transform_points4_identity() function.  Use the x86 version instead
      regenerated to support new 1.5 functions, etc.
      Use same steps for Linux and OpenBSD.
      added aix-gcc config
      added OSMESA_LIB_DEPS needed for some platforms
      removed MAKELIB lines.  tweak openbsd config
      aix-gcc tweaks
      more aix-gcc tweaking
      add GL_TEXTURE_CUBE_MAP_ARB to _mesa_max_texture_levels(), reordered some comments
      call _mesa_delete_texture_object() from in the driver's DeleteTexture function
      fix vertex program debugger issues (Bug 874382)
      change -O to -O2 for aix-gcc
      comment-out the touch/rm depend lines
      make executable
      added note about mklib permissions
      latest bug fixes
      updated to version 21 with OpenGL 1.5 support
      remove stuff that's now in glext.h
      glDeleteProgramsARB() failed for fragment programs (bug 876160)
      exercise glDeleteProgramsARB
      updated glext.h, fixed bug 876160
      tweak aix-gcc config
      bump MAX_WIDTH/HEIGHT to 4K pixels
      latest changes, bug fixes
      added blurb about NVIDIA extensions
      use -G not -shared for gcc on SunOS
      change MALLOC to CALLOC to silence valgrind error in _playback_copy_to_current()
      Cosmetic changes.
      fix -fomit-frame_pointer typo
      tweaks for 6.0 release
      bump version to 6.1 (new development)
      added some missing FLUSH_VERTICES calls
      tweak openbsd config
      tweak OpenBSD and SunOS commands
      make second param of driParseConfigFiles() const
      include sched.h to get sched_yield() prototype
      s/char/GLubyte/ to silence warnings
      include texobj.h to silence warnings
      change Data ptr from GLbyte to GLubyte to silence warnings
      Change get_ust_nop()'s parameter from int64_t to uint64_t to silence warnings.
      minor changes to silence warnings
      fix glitch from previous check-in (pointer vs int)
      omit CVS directories from tarballs, bump version to 6.1
      remove incorrect comments, added _mesa_bzero() call in _mesa_initialize_texture_object()
      replace CALLOC with MALLOC in _mesa_new_texture_object()
      replace -lm with $(GL_LIB_DEPS) to get -lexpat
      _mesa_init_driver_functions() to initialize dd_function_table
      Before calling _mesa_create_context(), initialize a dd_function_table struct
      initial 6.1 release notes with info about recent _mesa_create_context changes, etc.
      Undo some bits from last check-in related to the ctx->Driver.NewTextureObject
      remove _mesa_check_driver_hooks() - it's really not too useful anymore
      don't call glVertex functions directly, call _glapi_Dispatch->Vertex(), etc
      misc updates
      Initial support for floating point and signed texture formats.
      fix assorted compilation issues
      a few more tweaks
      more fixing for tdfxUpdateTexturePalette
      Change software alpha plane pointers from void* to GLchan*, eliminate some casts.
      added device driver hooks for BindProgram, NewProgram, DeleteProgram
      added link to ORSA (per request)
      New glXUseRotatedXFontMESA() function.  Like glXUseXFont(), but takes
      initial info for version 6.1
      Some initial RGB and RGBA floating point texture formats.
      New error checking in _mesa_GetTexImage().
      updated GLUT link
      replace MALLOC w/ CALLOC to silence valgrind warnings
      remove bogus assertion in DeleteTexture() function
      change PFNGLXGETUSTPROC's argument from uint64_t to int64_t to match the DRI
      s/BlendEquatioRGB/BlendEquationRGB/
      patch from bug 885992
      minor fixes to glw's touch depend
      tweak the #if test for defining _ASMAPI (fixes IRIX problem)
      fix C++ link issues on IRIX
      fix ARCHOPT typo
      more IRIX tweaks
      replace spaces w/ tabs
      more descriptive error strings
      Via Unichrome/cle266 driver (Erdi Chen)
      Minor fixes for recent Mesa 6.1 changes
      added unichrome directory
      glBindProgramARB wasn't getting compiled into display lists (bug 887383)
      add missing CXX, CXXFLAGS for osf1 (Bug 886628)
      don't call gl functions directly, jump through _glapi_Dispatch table
      enable program and arrays by default, minor clean-ups
      add vparray.c, fix indentation
      added _NEW_PROGRAM to check_state flags for pipeline stages (fixes vparray demo bug)
      minor #ifdef fix
      added display list option (l key)
      tweak comments
      glXChooseFBConfig() didn't handle GLX_FBCONFIG_ID option
      updated list of env vars
      s/getenv/_mesa_getenv/
      another pbuffer demo (Andrew Lentvorski)
      added Daniel Stone for getting CVS access
      fix two-sided lighting / vertex program bug (#887330)
      another tweak to two-sided detection test
      fix stores to vertex state program registers
      fix typo
      fix glXChooseFBConfig bug 890167
      bring in Keith's _math_matrix_ortho() compiler work-around
      update function mangling and add comment about it to gl.h
      added GLU_INCOMPATIBLE_GL_VERSION
      Added flags for _all_ extensions to the gl_extensions struct so that drivers
      implement Fake_glXCreateWindow() to fix bug #890894
      added an additional comment about mipmap generation
      fully parameterize the macros for fixed-point arithmetic
      Updated to handle generic vertex attributes accessed vi glArrayElement().
      tweak solaris-x86-gcc and ggi targets
      Restore some of the previous code for handling conventional vertex attributes
      some useful matrix functions
      Do more bookkeeping of vertex buffer object reference counts.
      Fix the problem found with UT after I had added support for glVertexAttrib.
      bunch of assorted updates/changes
      Minor re-org of state for fog and precision options.
      new comments, rename some stuff
      set buffer object default Usage and Access
      init secondary color to (0,0,0,1).  remove some redundant initializations.
      info about precompiled libs for Solaris
      added note about kernel modules
      Minor clean-ups of variable scopes, initializations, etc.
      some code clean-ups
      remove incorrect negation in fog coord computation
      Added TRIANGLE_WALK_DOUBLE to optionally walk triangle edges with GLdoubles
      added FixedToDouble() macro
      added FogOption to fragment_program struct
      some debug code
      just some reformatting
      added SPAN_W
      define DO_FRAGMENT_FOG=1 to test fragment program fog option
      A bit of an overhaul of the fog code.
      compute fog coord in eye space, not clip space
      A few more tweaks to fog code.
      recent bug fixes/changes (fog, glArrayElement, triangle rasterization)
      some minor tweaks
      fix typo (Matt Case)
      updates from Erdi Chen
      glDrawArrays loop was wrong
      added some const keywords
      Yet another revision of _ae_update_state().  Issue both conventional AND
      upgraded to version 22
      s/transform/tnl/
      added some GLAPIENTRY keywords, minor clean-ups
      fix incorrect texcoord attribute index
      jump through dispatch table instead of calling gl functions directly
      turns out we probably need a _tnl_allow_pixel_fog() function afterall
      call _tnl_allow_pixel/vertex_fog() funcs
      use ABS(Z) when computing blend factors
      added _tnl_allow_vertex/pixel_fog() calls
      Fix C/C++ calling conventions problem (Bug 904854)
      fix library naming glitches (bug 904867)
      fix another glDrawArrays indexing bug, jump through _glapi_Dispatch functions
      check for invalid vertex/fragment program in glBegin
      check for invalid fragment program in glDrawPixels, glCopyPixels and glBitmap
      check for null paramList
      fix PK2H, UP2H, PK2US bugs
      move _mesa_load_state_parameters() to state validation stage
      paltex demo from recent texcombine changes.  Minor clean-ups
      replace color table FloatTable boolean with Type enum
      consolidate some code
      move _swrast_texture_table_lookup() to _mesa_lookup_rgba_chan()
      rename some span pack/unpack functions for better uniformity
      consolidate image transfer operations in new _mesa_apply_rgba_transfer_ops() function
      minor clean-ups
      Remove clamp parameter from _mesa_unpack_color_span_float().  Pass the
      added missing error check in _mesa_BeginQueryARB (Cedric Gautier).  minor clean-ups.
      fixed linux-osmesa16-static.  Added linux-static config
      more changes to VBO reference counting and deletion
      s/_swrast_exec_nv_fragment_program/_swrast_exec_fragment_program/
      comments, capitalization, misc-clean-ups
      added clamping to polygon offset to prevent potential negative Z values and FP exceptions
      yet another take on VBO deleting and reference counting
      silence warnings
      casts to silence warnings
      remove some unneeded macros
      a bunch of assorted clean-ups, etc to fix warnings, etc
      plug in _mesa_BlendFunc into dispatch table
      remove redundant macros (also in m_matrix.h)
      Fix color index mode rendering.
      remove unneeded casts
      don't futz with GL_CLIENT_ACTIVE_TEXTURE in glClientActiveTexture (Robert Merrill)
      more descriptive error messages for matrix stack over/underflows
      Implementation of GL_EXT_pixel_buffer_object extension.
      Added some comments.  Minor const, int->GLint type changes, etc.
      GL_EXT_pixel_buffer_object test
      fix stride bug in RGB and BGR span rendering
      more TRIANGLE_WALK_DOUBLE tweaks (Justin Novosad)
      assorted updates and new AIX support (Dan Schikore)
      additional configs (Dan Schikore)
      remove redundant call to ctx->Driver.Viewport(), and clean-ups
      add missing else statements
      make null-terminated copy of program string so that parse works correctly
      minor clean-ups
      fix array index error in _swsetup_Translate (Felix)
      Implemented support for software-based AUX color buffers.
      test of aux buffer rendering
      added aux buffers item
      added s_auxbuffer.c
      use GL state vars instead of program parameters
      added some code to test buffer deletion
      new DRI interface header
      additional updates from Dan Schikore
      tweaks to Darwin code
      rename class to c_class if using C++ (bug 922294)
      use Driver.WriteMonoRGBAPixels() when possible (Dan Schikore)
      update my email address
      added linux-x86-opteron (bug 922735)
      silence warnings
      re-order arguments to mklib
      more Darwin tweaks
      merge changes from 6.0.1 branch
      check for GLX 1.3 during init
      tweaks for 64-bit linux configs (Dan Schikore)
      new system configurations
      New Makefile system
      Mesa installation script
      updated Makefile info
      omit CVS from configs list
      added OSMESA_DRIVER_OBJECTS and COMMON_DRIVER_OBJECTS
      added _DEPS lines to specify -L/usr/X11R6/lib
      added glw to SRC_DIRS
      obsolete
      clamp BiasMin, BiasMax to reasonable values
      increase MAX_TEXTURE_LOD_BIAS to 11.0
      Tweaks for make clean / realclean.
      new Makefile
      Don't use pushd/popd.
      don't try to make GLwMDrawA.c by default since it needs Motif
      added .SUFFIXES : .cc line
      tweaks to APP_LIB_DEPS (Dan Schikore)
      fix some warnings
      added GLW_SOURCES for IRIX
      minor fixes
      fix the build
      Take a stab at updating the VMS files.
      fix array indexing for UP4B and UP4UB
      merge from 6.0.1 branch
      added hpux10-static config
      Insert original fragment program instructions as comments.
      Reorder some code so we don't upset program state when there's a parsing error.
      these files now live in the shader directory
      these are now in the shader/ directory too
      fix error in rgb332 texel fetch function
      added FLUSH_CURRENT() call in GetMaterial() calls
      add braces to silence warnings
      Implement ARB_f_p KIL correctly.
      fix bad array indexing in _save_current_init() that caused context state to get clobbered
      added some assertions, just to be safe
      merge from 6.0 branch
      reorder 6.0 / 6.0.1 entries
      shouldn't be in CVS
      for make clean, include tests directory
      minor tweak
      minor corrections
      update revision history
      use # comments instead of //, etc (ajax at nwnk.net)
      use new MKDEP, MKDEP_OPTIONS variables
      move 'id' declaration so it's C compliant
      don't include GL/gl.h, rely on glheader.h
      no depend file
      move pSAREAPriv declaration to be C-compliant
      lots of Makefile updates for new build system
      minor fix
      added -noprefix option for making DRI drivers
      use MKDEP var
      added zreaddraw.c
      added SPECULAR flag for sprite point functions
      added extern "C" for C++ (bug 930933)
      fix max buffer size test (Kyle Bateman, bug 912828)
      fix minor glitch with GL_DEPTH_COMPONENT textures
      check for __AMD64__ when setting USE_IEEE (Ronnny Vindenes)
      minor tweak for Darwin
      set CONFIG_NAME = linux-tcc
      wrap file in #ifdef USE_TCC, otherwise compilation failes
      Audit/fixes for NV/ARB TEX, TXP, TXB, TXD instructions.
      hack to use glTexCoord4f and Q variable to test projective texcoords
      added missing SHADER_SOURCES to ALL_SOURCES
      updated instructions for new directory tree and makefile system
      check for __ia64__ when setting USE_IEEE
      bug fixes for ARB/NV fragment program texture instructions
      do texcoord projective division in _swrast_span_default_texcoords()
      in TexEnv functions, check if pname == GL_TEXTURE_ENV_COLOR and fetch appropriate number of param values
      glTexEnviv bug fix
      linux-solo-ia64 config (Stephane Marchesin)
      apply texgen/texture matrix in glRasterPos
      removed non-existant arbfpwpos.c, added UTIL_FILES target
      check return values of _swrast_CreateContext, etc
      only init temp regs to zero if executing NV fragment program
      Fix up some assorted issues with initialization of vertex program registers.
      New glTexImage code.
      obsolete
      minor clean-ups
      fix some include file glitches
      use FetchTexelf() in GetTexImage() to return float textures correctly
      added floattex.c to test floating point textures
      added CYGWIN support
      s,CHAN_BITS/8,sizeof(GLchan),
      new CYGWIN docs from Eric Lassauge
      remove some errant code from previous check-in
      assorted fixes
      Added ctx->Vertex/FragmentProgram._Enable flags.  Set when vertex/fragment
      init mgaScreen->irq (Tilman Sauerbeck)
      GL_DOUBLE doesn't follow GL_FLOAT, fix indexing (Keith Harrison)
      move #define for GL_DOUBLE, to put enums in right order
      minor fixes for entrypoint name mangling
      call drmUnmapBufs() in mgaDestroyScreen (Tilman Sauerbeck)
      fix HP-UX typo (bug 941576)
      s/-mv8/-mcpu=ultrasparc/ (bug 941338)
      added pointer to Vincent (OpenGL ES implementation)
      Removed the old teximage code.
      updated other makefiles
      minor vertex/fragment program tweaks
      restored _mesa_rescale_teximage2d()
      fix StoreImage calls, remove dead code (Adam Jackson)
      Fix minor warnings found with g++.
      remove unneeded line
      obsolete
      fix a bug on the convolution path
      added -c option to exercise convolution of glTexImage2D
      documentation for TexImage code changes
      fix a few bugs when using GL_REDUCE convolution mode
      don't need auto mipmaps
      Added menu options to test border modes and linear/nearest filtering.
      fix rectangle texture clamping and border-related code
      s/GLX_PIXMAP_BIT_SGIX/GLX_PBUFFER_BIT_SGIX/
      Don't call calculate_derived_texenv() from in _mesa_TexEnv().
      Added big-endian texture formats.
      yank bgr233 texformat.  minor comment updates.
      obsolete in favor of regular Makefiles
      fix red/green/blueBits typos
      remove the last line of the previous patch
      remove last remnants of NEWTEXSTORE stuff
      fix parameters to StoreImage calls
      fix compiler warnings (Jerome Glisse)
      some component ordering bugs in extract_float_rgba()
      GL_BGR can't be used with the packed types, according to the GL spec.  Enforce such.
      Test program to check that packed pixel formats work as expected with
      Another pass at implementing byte-swapped texture formats.
      fix parameters to StoreImage()
      silence warning in assignment to gp.value
      cast driHwLock assignment
      Minor fixes/optimizations to type conversions in draw_depth_pixels().
      minor doxygen updates
      Replaced 'core' with 'main'.
      fixed some bogus array enable tests (Soju Matsumoto)
      updated URL
      lots of updates
      s/core/main/
      remove duplicate GLX_DRAWABLE_TYPE token, sorted tokens by enum value
      fix bug #957792 (swap buffers), plus misc clean-ups
      only use simple_z_textured_triangle if depthBits <= 16
      check for null pointer in Fake_glXCreateGLXPbufferSGIGIX() (bug 961376)
      Check for NULL texture object when choosing texture sampler.  Fixes segfault when fragment program references an incomplete texture
      fix stencil index bug (#964704)
      fix parse error in sign of exponents (bug 968323)
      Roland's patch to fix promotion of alpha or luminance textures to the
      Additional error checking for glTexImageXD.
      document recent changes/fixes
      don't pass GL_INTENSITY for <format> to glTexImage2D - it's illegal
      fix minor typo in comment
      fix some minor glitches in glGetTexParameter code
      added GL_BGR and GL_BGRA to is_color_format() (Ronny Vindenes
      remove the 3dfx CatchSignals stuff (the var was alwasy set)
      s/_mesa_get_current_context()/GET_CURRENT_CONTEXT()/
      disable GL_INTENTSITY in _mesa_is_legal_format_and_type().  See table 3.6 of the 1.5 spec
      just a comment
      added compressed formats to is_color_format() and updated some comments
      don't use // comments (Thomas Hellstrom)
      fix component indexing bug in _mesa_texstore_al88()
      Remove ADD_POINTERS code from import() function - it broke UT2004.
      Big-endian texture fixes from Michel Dänzer.
      store's to RC/HC didn't work (bug 976287)
      latest bug fixes, etc.
      remove redundant program field initializers in BindProgram(); ctx->Driver.NewProgram() does the inits
      call ctx->Driver.NewProgram() instead of CALLOC/etc in _mesa_LoadProgramNV.  Fixes problems found with Cg demos.
      check for NULL pointer to glTexImage (fix verified by Adam Jackson)
      Alexander Gottwald's patch for cygwin
      updated to version 23
      Fixed occlusion_zless_triangle() to work with 16 or 32-bit Z buffers.
      added library dependencies (bug #978758)
      fix capitalization, new comments, etc
      fix HP occlusion testing when using generic span path
      fix GL_HP_occlusion_test bug, use glext.h version 23
      remove program from hash table when deleted (bug 979514)
      make sure Q has a good value before drawing a sample
      glBegin/EndQueryARB didn't work inside display lists
      glBegin/EndQueryARB bug fix
      option to test with display lists
      remove trailing whitespace and misc clean-ups
      disable some debug code
      the get_register_pointer() function was pretty well broken in a few ways
      fix another bug in get_register_pointer()
      fixed vertex program state references
      added aix-static config, make aix config build dynamic libs
      added commands for AIX shared libs (Shane Blackett)
      just a test; no change
      just a test; no change
      just a test; no change
      just a test; no change
      just a test; no change
      init some vars to silence warnings
      redo the event loop to make event response better
      disable some debug printfs
      a few comments
      document some assumptions made about compressed texture tile size
      added cast to fix AIX color interpolation problem (bug 983144)
      patch for AIX (bug 983144)
      fixed AIX color interpolation bug
      remove -lXm (motif) from GLW_LIB_DEPS since it's not needed by default
      new image (see mesa bug #993926)
      fix compare w/ zero warnings (bug #988766)
      error message fixups and some debug code
      fix compare w/ zero warnings (bug 988766)
      ALIGN16 macro repairs
      added support for MESA_GLX_FORCE_ALPHA env var
      added MESA_GLX_FORCE_ALPHA env var
      updates from Stephane Marchesin
      minor comments
      new MESA_GLX_DEPTH_BITS env var
      use GL_CALL() macro (Andreas Stenglein)
      consistantly use unpack parameter, instead of ctx->Unpack (Daniel Borca)
      don't use GLuint or GLvoid in function declaration
      version 24 from oss.sgi.com
      version 6 from oss.sgi.com
      vtk and Paraview links updated
      added a few more debug printfs (disabled)
      added some handy debug code (disabled)
      query/print max convolution filter size
      minor tweaks
      use 'a' to toggle animation
      silence warning
      minor anim tweak
      updated glext.h and glxext.h files
      cvs ignore file
      added some missing sources to tarball list
      minor fix for newest glxext.h header
      silence a warning
      fix some memory leaks (bug #1002030)
      indentation fixes
      BeOS updates, memory leak fix, updated release notes
      minor update to intro, and some word wrapping
      updated tarball list for BeOS changes
      fix swrast/shader typo
      whitespace clean-up, new 'clean' commands
      some updated info about adding new extensions
      don't enable the float texture and GLhalf pixel extensions yet
      set release date for 6.1
      updates from Jouk
      check for either GL_NV_vertex_program or GL_ARB_vertex_program for some enables
      added Mesa 6.1 release info
      don't check for ARB_texture_non_power_of_two for OpenGL 1.5 (Philipp Klaus Krause)
      bump version to 6.2
      first bug fix for v6.2
      enabled GL_ARB_texture_rectangle
      also check for __hppa__ and hpux for IEEE support (bug 1003026)
      check if __mips or _MIPS_ARCH is defined to set IEEE_ONE (bug 1003026)
      don't use __FUNCTION__ - not portable
      cast to fix a warning (in disabled code anyway)
      added src/glw/depend to tarball list
      IRIX fixes
      release notes for 6.2
      work around a param bug (Adam Jackson)
      print current config name before building src and progs
      remove BeOS conditional code.  oglinfo should be moved to xdemos since it's a GLX program
      fix a couple bad casts
      Windows ICD driver code (patch 1014800)
      Fix mgl name mangling (patch 1014800)
      return type for MapBuffer was not specified
      fix MGL namespace warning (patch 1014800)
      use version generated from the XML database
      Makefile for python-generated files (Ian Romanick)
      regenerated with license info
      Silence gcc 3.4 warnings on ReactOS.  Mostly unused var warnings.  (patch 1015696)
      print internalFormat value in r200ChooseTextureFormat() error message
      move (void) foo; instances after local declarations
      ICD update from Gregor Anich
      bump version to 6.2, add installmesa to tarballs
      updated CFLAGS (Anich Gregor)
      also test for __arm__ for setting USE_IEEE (X bug 1155)
      Another attempt at fixing name mangling and gl__unused413().
      initial bits for DRI_FILES for putting DRI driver sources into the tarball
      clean-ups and comments
      fix a bug in analyse_from_scratch() reported by Wes Bethel
      added glapi and shader modules, misc updates
      More updates for Doxygen.
      reset error state before parsing (David Reveman)
      fix matrix classification and program parsing error state bugs
      updated error msg
      misc updates for 6.2 release
      assorted updates (bug #1025977)
      ggi updates
      removed (bug 1026109)
      new file (bug 1026109)
      updated from patch 1026109
      applied patch #1026109
      updated ggi files in tarball list
      tweak texcoord for sampling texture rectangles (Dave Reveman)
      adjust texcoords for texture rectangle sampling
      remove ARB_texture_non_power_of_two from _mesa_enable_1_5_extensions()
      Handle GL_MAX_VERTEX_ATTRIBS_ARB in glGet*().
      added GL_MAX_VERTEX_ATTRIBS_ARB bug fix
      Replace -lGL with -l$(GL_LIB), etc.
      Repeatedly deleting a texture ID with glDeleteTextures() could lead to a crash.
      Have the rendering contexts share textures.
      fixed glDeleteTextures bug
      also fix possible delete bugs with buffer objects and vertex/fragment programs
      in _mesa_problem() include version number in error message
      replace glNormal3f(v) with GL_CALL(Normal3f)(v), etc (Andreas Stenglein
      clean-up the hokey transformation code so that window resizes actually work
      VBO refcount and spriteblast fixes
      silence warnings, re-indent code
      silence warnings
      just some comments and formatting changes
      added a comment in BindTexture
      removed hacked GL tokens, fixed a printf
      don't test for NULL pixels pointer here, do that in the 'store' routines after validating PBO address
      patches for clearing hw depth buffers from software fallback (Nicolai Haehnle)
      updated osmesa info
      moved in from the tests/ directory to include in distro
      moved to demos/ directory
      VBO RefCount fix (David Reveman)
      bump minor version to 2
      remove beos/*.h files from tarball list
      change include of glx_mangle.h to match other headers
      version bump
      set span.y prior to each span write in draw_depth_pixels, like Karl's other fix
      set 6.2 release date
      added PBO known issues
      fix LoadProgramNV regression when I had fixed the RefCount bug
      a few more updates for the 6.2 release
      initial stuff for 6.3
      bump version to 6.3
      fix a comment
      fix indentation
      added support for GL_ARB_draw_buffers
      fix _DrawDestMask[0][0] typos
      GL_ARB_draw_buffers
      remove unused413 stuff, glDrawBuffersARB uses that slot now
      put glapi_x86.S into the x86/ directory
      ARB_fp support for GL_ARB_draw_buffers (Karl Rasche)
      make clean should remove depend.bak too (Sérgio Moneiro Basto)
      fix GetVertexAttrib problem
      more GL_ARB_draw_buffers updates from Michale Krol
      Fix minor fog / fragment program state bug.
      initial 6.3 changes
      updated fsv URL
      fix if/else bug in _mesa_ProgramEnvParameter4fARB (fdo bug 1645)
      if CXX or CC env vars aren't set, use g++, gcc by default (Linux)
      GL_OES_read_format info
      added initial support for -exports option, Linux/OpenBSD only for now
      minor fixes from J.P. Delport
      addes osdemo32 target (J.P. Delport)
      wrap angle value (Philipp Klaus Krause)
      added stereo option (Jacek Rosik)
      added an attention-getting comment for developers who get stopped in _mesa_test_os_sse_support() while debugging
      updated PLIB URL
      name parameter to glGetActiveAttribARB() should not be const.
      minor clean-ups, found with pychecker
      obsolete
      VertexAttrib3svNV was incorrect
      Updated getprocaddress test to use gl_API.xml instead of old APIspec file.
      remove debug code
      Remove redundant error checking.
      use _mesa_unmap_buffer by default
      unbind GL_PIXEL_UNPACK_BUFFER_EXT before calling glBitmap!
      Allow the software fallback glDrawPixels, glReadPixels, glBitmap commands to
      Removed _swrast_validate_pbo_access().
      Map/Unmap PBO as needed so that the texstore routines can work with
      glGetTexImage() now works with PBOs.
      PBO support for glGetCompressedTexImage().
      glPolygonStipple() and glGetPolygonStipple() updated for PBOs.
      updated for PBOs
      PBO support for glConvolutionFilter1D/2D, glGetConvolutionFilter, etc.
      PBO support for glColorTable, glColorSubTable, glGetColorTable, etc.
      PBO support for gl[Get]PixelMap functions
      Use the _mesa_scale_and_bias_rgba() function in the convolution functions.
      indentation fix
      finished up PBO support
      remove PBO work from to-do list
      added a cast to ADD_POINTERS macro
      don't advertise GL_ARB_vertex/fragment_shader until they're done
      fix convolution regression from revision 1.48 to 1.49
      merge the AIX and AIX64 cases
      shorter error messages when calling a GL function without a bound context (FDO bug 1775)
      Remove need for defining _MSC_VER when building Mesa for windows with
      don't use ctx->Pack, use the function argument
      added _mesa_clip_drawpixels() and _mesa_clip_readpixels()
      Removed _swrast_clip_pixelrect().  Use _mesa_clip_drawpixels() instead.
      added a comment and assertion in _mesa_clip_drawpixels() for PixelZoom
      use the new _mesa_clip_readpixels() routine
      do clipping prior to XGetImage, just in case the image would extend beyond the screen's bounds
      added shaderobjects.c
      GL_(UN)PACK_SKIP_IMAGES should only be applied to 3D texture pack/unpacking
      allow specifying image file on command line
      fix APP_LIB_DEPS (bug 1065260)
      misc updates
      include <X11/Xthreads.h> instead of "Xthreads.h"
      applied SPARC patch (freedesktop bug 1898)
      silence warning, minor clean-up
      added GL_RED/GREEN/BLUE to is_color_format()
      Don't allow queries/etc of buffer object 0 - it's invisible to users.
      update _mesa_store_teximageXd() comments and minor code clean-up
      minor APIENTRY clean-ups
      Started some assorted clean-ups in #defines, typedefs, etc.
      Some new comments, clean-up formatting, etc.
      Use __GLXextFuncPtr type instead of void * for generic functions to
      Use a generic function typedef instead of void * to avoid gcc 3.4 warnings.
      update some type names
      (GLint*) casts in glGetIntegerv calls
      added X11 and OSMESA source to ALL_SOURCES, updated etags command
      Removed GLCALLBACK stuff - apparently never used anywhere.
      removed a bunch of #define cruft, in case anyone still uses this code
      remove the GLAPI/GLAPIENTRY defines - they're already in gl.h
      fix typo, update version/date
      Use new _glapi_proc typedef instead of void * for function pointers.
      use _glapi_proc
      make get_static_proc_address() return a _glapi_proc
      use @ to silence some unneeded output
      remove -Wmissing-prototypes from g++ flags, per gcc 3.4
      typedef GLUTproc, return it from glutGetProcAddress()
      clean up 'depend' commands
      clean up code related to dispatch table initialization
      Have OSMesaGetProcAddress() return new OSMESAproc typedef.
      new GetOverlayInfo() function to reduce code and silence warnings
      don't need glut_fbc.c
      use new PUBLIC macro for symbol export
      Set symbol visibility to 'default', if supported by gcc.
      misc updates
      silence warnings
      Change the dispatch offsets for the VertexAttrib*NV functions so they don't
      add a few functions in add_newer_entrypoints()
      remove _glapi_add_entrypoint() calls, they're already in context.c
      Remove the redundant width, height fields in xmesa_buffer struct.  Just use
      new comments and assertions
      Remove _mesa_ResizeBuffersMESA() call from _mesa_set_viewport().
      fix missing width/height error
      update from Jouk
      fixed OSF/1 shared lib problem (bug 1065260)
      added __glXFindDRIScreen to glXGetProcAddress (bug 1068879)
      silence a warning
      no need to include stddef.h anymore
      silence a variety of warnings found with g++ 3.4.2
      re-disable TRACE
      silence warning
      Use the GL datatypes.  Lots of assorted clean-ups.
      mask color indexes against palette size, per the spec
      direct bug reports to freedesktop.org
      report bugs at freedesktop.org
      check if using a PBuffer in clip_for_xgetimage()
      bring in 6.2.1 changes
      typo
      Added driver hooks for GetTexImage() and GetCompressedTexImage().
      fix recursion problem introduced on Nov 27
      removed unused visInfo var
      s/DP3/DP3_SAT/ to prevent negative values
      tweaks to the LIT instructions
      init span.w, dwdx, dwdy in case we're running a fragment program
      fixes for fragment programs
      don't divide texcoords by q if using a fragment program
      Experimental PRINT instruction for NV_vertex_program.
      fix typos in texcoord setup - fixes perspective correction bug
      VMS update
      Added PRINT instruction for GL_NV_fragment_program.
      added GL_ARB_pixel_buffer_object extension string
      Consolidate _swrast_write_texture_span() into _swrast_write_rgba_span().
      fix bug in _mesa_IsTexture()
      updates from Jouk
      DirectFB driver (Claudio Ciccani)
      bump version to 6.3, etc.
      latest updates
      Use the GLX 1.3 pbuffer/fbconfig functions in preference to the SGIX ones.
      directfb changes (Claudio Ciccani)
      use HIDDEN macro to export fewer symbols (bug 2210)
      added GLX_NV_float_buffer
      In choose_visual() we need to know if we're being called from
      New GetAllFBConfigs() function.  Some FBConfigs weren't reported before.
      added GetAllFBConfigs()
      fix-up rules for sphere demo, which uses readtex utility
      compute reasonable animate rate (Marcelo Magallon)
      rotate at a reasonable rate
      animation rate patch (Marcello Magallon)
      Add #ifdefs for extension testing (Marcello Magallon)
      include readtex.h instead of readtex.c (Marcello Magallon)
      include readtex.h instead of readtex.c
      some new/improved rules
      better animate rate (Marcelo Magallon)
      include readtex.h
      use #ifdef to test for extension
      rotate at fixed rate (Marcelo Magallon)
      rotate at fixed rate, changed fire effect (Marcelo Magallon)
      improved animation rate
      minor readtex fixes
      Windows/ReactOS patch (Gregor Anich)
      tweak output formattting, fixed uninitialized var error
      remove a debug line
      VMS updates from Jouk Jansen
      prototype _mesa_init_ati_fragment_shader()
      Some initial work for OpenGL 2.0: glStencilFunc/Op/MaskSeparate() functions.
      OpenGL 2.0 StencilFunc/Op/MaskSeparate functions
      temporary prototypes for OpenGL 2.0
      make tnl_clipspace_attr->insert field const, to silence warning
      _mesa_GetInteger/Float/Boolean/Doublev() are now generated with the new
      added a few OpenGL 2.0 tokens (temporary)
      added getstring.c
      fix a few warnings
      silence a bunch of warnings
      clean up gl2 structs
      silence warning
      fix some weirdness accidentally checked in in the past
      use FABSF(fogCoord) to fix fog calculation problem (Soju Matsumoto)
      added rule for compiling C++ sources
      minor reorg of shader files, disable building for now
      minor formatting fix
      fix a typo
      make sure the new instruction Data pointer is set to NULL
      Change behaviour of glDeleteTextures as discussed on ARB list.
      test program for texture sharing, deleting, etc
      added texobjshare.c
      Removed gl_texture_object's DeletePending.  Changed a comment.
      Use a simple CSS style sheet.
      Update glDeletePrograms/Buffers() so that the ID is freed immediately, like
      document new glDeleteTextures/Buffers/Programs behaviour
      glext.h version 26
      fix glGetString funniness
      Added _mesa_HashNextEntry() function to allow walking over all entries
      reenable some assertions
      Use the _mesa_HashNextEntry() function to walk over all textures in the
      just fix a comment and update date
      added mutex to-do comment
      just comments/clean-up
      remove a bunch of temporary #defines, now that glext.h was updated
      added missing glFinish()
      fix initial viewport size bug
      assorted fixes
      added osmesa16 stuff
      remove stray span.y++
      fix max viewport query
      clamp viewport against ctx->Const.MaxViewportWidth/Height instead of MAX_WIDTH/HEIGHT
      removed need for Shared->TexObjectList, walk over texture hash table entries instead
      remove unneeded code
      remove dependency on Shared->TexObjectList
      add a few comments
      Remove the Shared->TexObjectList pointer and Next field from gl_texture_object.
      Some initial work on GL_EXT_framebuffer_object.
      added fbobject.c to sources
      additional work on GL_EXT_framebuffer_object
      improved warning message
      indentation fix
      added missing prototype
      checkpoint latest work
      Add new entrypoints for GL_EXT_framebuffer_object.
      added GL_EXT_framebuffer_object enums/funcs.  Remove when glext.h is updated
      removed GL_EXT_framebuffer_object tokens
      plug in GL_EXT_framebuffer_object functions
      a bunch of assorted fixes
      just some comments
      clamp anisotropy against max value
      implement the 'completeness' tests
      initial support for GL_EXT_framebuffer_object
      Added a test/clamp in the scale_internal_*() functions to prevent reading
      glGet*(GL_FRAGMENT_PROGRAM_BINDING_NV) was returning the vertex program binding.
      lots of updates
      new comments, fix zoffset error test
      unlock mutex upon error return (Jeff Muizelaar)
      s/0/NULL/ (Jeff Muizelaar)
      fixed problems with parse_float() (fd.o bug 2520)
      fix non-Win32 definition of GLUTAPIENTRY, remove some cruft
      remove extern (Jeff Muizelaar)
      s/0/NULL/ (Jeff Muizelaar)
      s/0/NULL/ (Jeff Muizelaar)
      XMesaGetDepthBuffer() returned wrong bytesPerValue (Karl Schultz)
      generate enums.c file (Zack Rusin)
      now generated with Python script
      Sort the enums in the Python code, instead of at runtime. (Zack Rusin)
      minor changes to aid debugging, and a new comment
      More GL_EXT_framebuffer_object: rename some things, added device driver hooks.
      Rename _mesa_update_buffers() to _mesa_update_draw_buffer_bounds() and do
      silence warnings

Daniel Borca (276):
      Reorganized combiners. Added optimized span functions.
      FEATURE_ARB_vertex_buffer_object
      Minor MinGW fixes
      GameMode
      glide driver (wip)
      minor fixes + doc update
      cosmetics; translated asm routines for use with ASSYNTAX.H
      Napalm total immersion
      fixed some NASM opcodes; added segment overrides and temporary labels
      documentation update; also new makefiles for Win32 MesaFX build
      fixed a typo (array size)
      small fixes to accomodate Intel compiler/linker
      cosmetics and memos
      added a few missing GLAPIENTRYs to make MinGW compile again
      more MinGW fixes
      entrypoints... again
      texture compression
      changes to accomodate texture compression
      bugfixes wrt texture compression
      removed detritus
      accomodate decorated names
      Voodoo2 happiness
      small fixes wrt texture handling
      small fixes wrt texture handling
      NCC fixes and some cosmetics
      dirty hack to advertise vanilla extension names
      vanilla extensions disabled
      doc updates; GLUT timer additions; fixed compilation warnings
      small matters
      texture compression testbed
      more texture compression
      updated core makefiles to include arbparse
      accomodate vtx-0-2-branch
      fixed a bug in texgen (T coord for GL_REFLECTION_MAP_NV)
      got rid of __FUNCTION__ :(
      fixed prototypes for wglGammaRamp3DFX
      small bug wrt culling
      EXT_secondary_color and EXT_separate_specular via multipass
      DMesa: removed MGA2064W driver
      reverted TMUs for proper multitexturing
      Mingw3 prototypes
      no-op table on par with main dispatcher
      sanity check for vertex counter
      cleanups
      wgl interface
      teensy bit of optimization in interp code
      protect against broken apps
      temporary workaround for bad primitives in fx_render_vb_quads
      bugfix in texture swapping mechanism (multitex)
      removed unnecessary (now) sanity check
      updated makefiles
      envvars to control run-time
      keeping things safe: sanity check back
      Glide driver (windowed mode) fix for Voodoo2
      updated FX wrapper
      small fixes
      re-enabled trilinearness
      added support for textures bigger than HW can support (SW rescaling)
      fixed a bug in texture memory manager when in UMA mode
      _mesa_delete_texture_object and some other updates
      documentation update
      implemented glutTimerFunc
      fixed refresh rate control in Glide wrapper
      fixed a bug in VGA initialization routine
      accomodated new driver_func initialization
      accomodate FetchTexel and fix some memory leak bugs
      glutTimerFunc cleanup
      minimize the number of DIVs
      GL_TEXTURE_1D and other stories...
      fixed a bug in GL_NEAREST sampler
      added glutGetProcAddress in DMesaGLUT
      added NUL driver for DMesa
      removed MGA driver from DMesa
      deal with combine_ext refactor
      reorganized DMesa to fxMesa wrapper
      fixed back fxMesaGetCurrentContext to match MakeCurrent
      fog_coord demo
      cleanup and some WGL PIXELFORMAT issues
      fixed some typos
      workaround for crash-upon-exit in verbose mode
      made NASM opcodes consistent
      _tnl_allow_*_fog(...)
      Voodoo Rush fixes
      fxMesa: remove buffer_obj_ext, cos it breaks UT2004
      unfinished SOLO driver for TDFX
      small changes (version string)
      code cleanup heheh
      fine-tuning x86 gcc codegen
      codegen'ed versions of the 2nd level dispatch
      added codegen'ed choosers
      disabled naughty cheats and protected the choosers behind AllowCodeGen
      got rid of the notify routine as intermediate step
      protected all codegen behind AllowCodegen
      revived linux-glide build process
      removed wfx makefiles
      updated Glide driver documentation
      documentation updates
      accomodate newtexstore
      transfer routines for compressed formats (avoid SEGFAULT, also prepare for external packers)
      updated makefiles
      FXT1 preparations
      FXT1: getting warmer
      unbloat namespace
      texture compression: getting warmer
      documentation update
      fixed bugs in fxddtex.c
      minor fixes
      added some core-level extensions
      added GL_NV_blend_square
      doc update; minor changes
      FXT1 texture compression (initial draft)
      oops
      reworked FXT1
      masked out unsupported glBlendColor (some games check only the proc addr)
      small corrections
      corrected a bug wrt point_size
      Fixed a bug in vertex emitter (untile we move to t_vertex.c)
      fix correct parameters passed to StoreImage
      Hack for linear fogging. Added empty routines for WGL_ARB_pixel_format WGL_ARB_render_texture WGL_ARB_pbuffer
      added support for non-64bit compilers
      added back x86 entrypoints
      disabled unnecessary assertions in texture replicator
      hack for compressed textures with aspectratio greater than 8:1
      put back the correct dispatcher for non-threaded builds.
      added GL_EXT_fog_coord.
      fixed non-threaded codepath
      added GL_ARB_point_sprite.
      use tnl_emit_func, it's safer
      fixed vertex RGBA
      SGIS_generate_mipmap
      GL_MODULATE/GL_REPLACE multitex (wasteful, but apparently needed by some apps)
      minor corrections
      several new bugs
      trying vertex_program...
      GL_EXT_texture_env_add shouldn't be conditioned by multitex, or should it?
      fixed a bug in clear routine
      minor doc update
      partial GL_BLEND for Voodoo1
      fixed pixelpiped DrawBitmap / DrawPixels (clipping was upside-down)
      Rush is back on ZBUFFER
      disable W1 when single-texturing
      fixed conflict between depth & alpha in 16bpp
      disabled automatic mipmaps until we can handle 565 textures.
      disabled stereo checking in wglChoosePixelFormat
      fixed pixel drawing routines, some cleanup
      fixed texture fetchers, fixed automatic mipmap generation, some cleanup
      cleanup
      prepared for release
      silence compiler warnings
      cleanup
      "const int _nc" was really stupid...
      why is GL_ARB_texture_env_combine different from EXT?
      cleanup
      hide GL_ARB_texture_compression from non-VSA cards
      Werror is bad, unless you know what you are doing
      remove CR
      fog really needs W
      fixed automatic mipmap generation
      clamp lodbias
      cleaned up extensions a bit
      revived unused code, but got rid of warnings
      cleanup
      remove CR
      document MESA_FX_NOSNAP
      newer Glide3x snaps vertices internally for V1/VR/VB
      make sure the HW is clean before shooting the pipeline
      changes towards GL_ARB_point_sprite (two-zero)
      cleanup
      cleanup and some changes towards GL_ARB_point_sprite (two-zero)
      cleanup and bugfixes for fxDDClear()
      clamp size of wide points
      added (back!) option to use non-packedcolor
      stencil wrap works, either HW or SW (Ian Romanick)
      use GLuint instead of uint
      enable 2_0 extensions
      stencil wrap works, either HW or SW (Ian Romanick)
      put Rush in the list of supported HW
      add newline to end of file
      add more mess to the dynamic linking system
      avoid the patch if Mesa_DXTn flag is set
      doc update
      bump version numbers to their real values
      SWTC trick
      SWTC trick, enabled new extensions
      SWTC trick (disabled for now)
      support for external libraries
      use RowStride instead of Width in texel fetchers.
      added dstRowStride to dxtCompressTexFuncExt;
      changed _mesa_rescale_teximage2d to allow rescaling of padded images
      small fix to match new  _mesa_rescale_teximage2d
      print an error message in convertPalette before dying.
      fix compilation error in `convertPalette' (still have to fix the issue, though)
      fixed compilation errors
      decoder "width" parameter represents "stride-in-pixels"
      make sure we don't generate random alpha
      fixed compilation error
      protected against old djgpp distros
      updated "clean" target in makefiles
      small aesthetic correction
      cleaned up the mess a bit
      pop(stencil) for EXT_stencil_two_side
      wip hack for EXT_stencil_two_side
      added DD_TRI_TWOSTENCIL to ease  EXT_stencil_two_side in device drivers
      added a few sanity checks
      small note regarding SSE under pure DOS
      emphasize on the contents of Glide SDK
      the win32 interface will be moving under windows/
      use parentheses to avoid warning
      use float constants
      fix _mesa_image_address2d invocation
      work around a bug in gcc 3.3.3 (which duplicates ASM block)
      moved windoze specific code outside drivers/glide/
      handle ENABLE_BIT for EXT_stencil_two_side
      sync with glut/glx
      small fix for debug paths
      disable junk for mingw
      explicit cast in a few places
      fixed an include path
      simplistic ICD implementation for fx/Mesa
      make sure we're cleaning up everything
      allow GetTexImage with RGBA format and COLOR_INDEX internalformat
      disabled ClipMask usage in vb emitter (clipping bugs?)
      protected elfish code against non-elf compilers
      added DMesaProc type to prevent gcc 3.4.3 warnings.
      added DMesaProc type to prevent gcc 3.4.3 warnings
      minor info update
      updated makefiles for gcc 3.4.3 and bnu 2.15
      fxRasterPrimitive wasn't always receiving the reduced primitive
      fxRasterPrimitive wasn't always receiving the reduced primitive
      really protect against npot compressed textures (logbase2 never returns -1).
      allow ARB vp/fp query program errors
      misc cleanup
      allow greater swappendingcount for older hw (v1/v2)
      comment reorg.
      made some functions static.
      comment reorg.
      removed detritus (ncc).
      removed ifdef'ed out code (to whom it may concern: it can be found in dri/tdfx driver)
      use float constants.
      put back clipmask code; it proved to be a MinGW/gcc 3.3.x error in t_vb_cliptmp.h(47), where it would still send clipped vertices to the hw... oh, dear...
      missing `or' in preprocessor conditional
      added GL_EXT_stencil_two_side (yes, it works)
      `t' was not initialized (use `texture' instead?)
      fix dstRGB factor
      added "default" branch in `convertPalette'
      removed detritus
      semantic in parameter names
      protected against elfish directives
      solved classic "char*" vs "char[]" conflict.  we were mimicking a pointer variable at desired location and then we took its address.  using array is more intuitive, as they give us the starting address instantly.
      semantic in parameter names
      semantic in parameter names.  changed emitter function.
      semantic in parameter names
      applied Keith's patch for "safe" type-punning.
      some "safer" type-punning (gcc optimizes float moves with integer moves, anyway).
      doc update
      changed coding style
      changed coding style.  added a NULL pointer check.
      changed coding style.  added possibility to use default timer instead of PC/HW timer.
      changed coding style.  made `glutTimerFunc' less accurate, but far more reliable.
      remove trailing slash from TOP definition
      since we've already messed up this one, let we do it further and make it look like a fire...
      fix compilation error for fx/Mesa
      allow more internalFormat/format combinations (i hope i got it right)
      cleaned up warnings
      There is no point trying to detect SSE at all, when MESA_NO_SSE is defined.  I am committing this because the current WIN32 SSE detection code crashes UnrealTournament2003.
      glVertex* is the provoking "cmd" (that is, by the time of glVertex*, we must have all other attributes already set).
      make sure we avoid assertion failure wrt VB->ColorPtr[1]->stride
      RGB/LUMINANCE texelfetchers (float version) returned 255.0F for alpha channel, instead of 1.0F
      makefile update (readtex, showbuffer)
      ProgramCallbackMesa must match the prototype
      `static const' instead of `const static'
      added a bunch of const in the decoder
      killed a few warnings
      separated DOS GLUT from GLX GLUT.
      removed detritus
      updated doc

Dave Airlie (101):
      fix breakage associated with pointers not being initialised in
      fix ugly bug triggered by macro expansion
      Initial mach64 driver import for DRI mach64-0-0-6 branch,
      Remove warnings from mach64 build, fix up some _SOLO stuff,
      fix missing include for SOLO build
      initialise VBLANK
      more updates , gears sorta works
      fixup vertex building code ..
      fix up specular color/fog and triangle vertex retrieval
      remove vertex_stride_shift, fix up code to not need it
      minor cleanups from t_dd_vb.c
      delete the textures
      back port span changes from i810, these have been in my tree for a while seem
      add xmlconfig.c to solo build
      use utility function to check versions
      Initial i810_dri.c for miniglx isn't fully functional yet...
      couple of steps closer still not fully working.. something amiss in ring
      extend SOLO code so the DDX version isn't checked
      remove version number faking, shouldn't be needed anymore
      Add includes to make mach64 build
      Add new interface support don't switch on yet ..
      more i810 to new interface
      make this exactly like XFree86 .. allows 4x to work
      add agpmode to miniglx config file
      make solo work like before new makefile merge
      Add shader sources for solo build
      get agpmode from config file
      Add agpmode to the example config file
      set default page flipping to zero can be changed in header file
      get the libraries the right way round..
      add drm source dir for solo includes
      make sure matypes is built
      DRM_SOURCE_PATH is in default
      fix up build for solo/dri
      get rid of last few XF86DRIClipRect
      clean up the common objects..
      add asm sources
      the config is included by the Makefiles that includes this file...
      add linux-dri-x86 target
      remove this use the one from the DRM
      switch to X86_SOURCES
      use X86 sources in X86 build, use none in other
      add r200 x86 sources
      stop crash when debugging
      fixup mach64 for newer build/types
      Remove use of X windows include
      make endian code work on BSD
      fix the versions
      oops fix a bug I introduced
      use the mesa defines
      fix all those warnings
      Fix rectangular textures on i830 - well texrect at least
      fix bug with emitting spec/fog report by Eric Anholt, this isn't the prettiest
      remove depend on clean
      make rm for common not complain
      fix mach64 non native.. no reason to use it...
      remove unused stuff, fix non-native build
      Add FXT1 compression support to i830 driver
      try and find the drm in a more likely place
      add i915 as it seems to build okay..
      switch to using driFillInModes fix depthbuffer = 0
      remove unused mode function
      Fix up texture compression mipmapping .. not sure if it fixes FXT1 but
      move over changes to intel driver from i830 driver for modes
      enable FXT1 texture compression for ixxx chipsets
      use a switch statement makes integrating the s3tc patch easier
      move config options to same place as other drivers (again for S3TC patch later)
      fix up heights for i830/i915 texture compression
      add debugging for compresssed textures
      xorg-r200-uninitialized-variable-used.patch from Redhat xorg
      bring over build fix from xorg tree
      bring over change from xorg trunk - let build on non-i386
      build fix from XFree86 tree
      Add a solo-x86 config, and make solo use the x86 glapi
      This patch enables GL_ARB_vertex_program and GL_NV_vertex_program
      updated patch for vertex program
      Add xmlconfig to the i810 makefile
      fix for Erics new emit state code, rework out the dest pointer
      Fix mesa solo with doublebuffered apps, the drawable index/stamp weren't
      From: Stephane Marchesin <marchesin at icps.u-strasbg.fr>
      fixup pci code for Jon's last checkin for changing page size
      back out Keiths last checkin - for Solo the include order is important
      Implement software ATI_fragment_shader
      add -DUSE_EXTERNAL_DXTN_LIB=1 so compressed textures are enabled for solo
      add sync debugging to the radeon driver
      reapply Keiths workaround for zbs emit that got dropped during
      add number of passes count to ati_fragment_shader
      fix signed vs unsigned comparison warnings in tnl_dd
      fix some signed vs unsigned warnings
      add EnableExtension so r200 enables all its extensions, add mesa memory
      add some missing texture formats and bits from r200 docs
      Add missing texture formats for completeness
      fix some 0->NULLs
      add more functionality to miniglx for glitz
      Add DESTDIR install target
      Add GLX_MESA_allocate_memory from Xorg glx.h
      bring over structs from Xorg glx.h to make Mesa glx.h the one true glx.h
      Add glXAllocateMemoryMESA demo app using ycbcr...
      use girl2.rgb by default
      Add a dri config option to enable the max texture level hack
      precedence precedence precedence.. add some brackets

David Bucciarelli (2):
      Updated to Voodoo driver v0.30
      Updated the DavidB email address and the MESA_GLX_FX related documentation

Erdi Chen (2):
      Fixed code that has computed incorrect number of modes.
      Corrected a float to ubyte conversion bug with fog enabled.

Eric Anholt (54):
      Remove usleeps from sis driver.  We probably aren't doing anyone a service
      Fix definitions for Intel compiler.
      Add support for Radeon IGP chipsets, based off of mcgrof-radeon-igp-v3.diff
      Don't mix variable definitions and other statements (fixed build on old
      Add missing sisInitDriverFuncs call.  Fixes segfault.
      Clean up some warnings by making sis_fatal_error a macro, and let it take an
      Use MKDEP make variables instead of hard-coding.
      Build fixing for FreeBSD.  GNU make is installed as gmake, so make a MAKE
      Missed in last commit:
      Remove unnecessary #include "mach64_common.h" which breaks the build for me.
      Add NEW_INTERFACE bits.
      Clean a little whitespace that's been bothering me.
      * Convert to use t_vertex.c instead of sis_vb.[ch]
      * sisChooseRenderState doesn't depend on _NEW_TEXTURE at all.
      Fix reflect demo: We were using the 3D Clear path unnecessarily, which appears
      Fix an "overflow in implicit constant conversion" warning in mEndPrimitive when
      Bugzilla #755: Fix polygon offset by setting DEPTH_SCALE to depth_scale instead
      Fix dri_util.c build with new headers.
      Remove unused HAVE_INDEX define.
      Close some races with locking on R100 and R200 which could manifest as rendering
      Revert the move of lost_context setting to UNLOCK_HARDWARE that was done in the
      Symbol names are prepended with an underscore on CYGWIN as well.
      The previous code would emit a full set of state during the first EmitState on
      Remove an unnecessary calculation of the dest pointer.
      Clean up some warnings in the R100 driver with -O -Wall.
      -O -Wall warnings cleanups in r200.
      Bump DRIVER_DATE for yesterday's changes.
      Fix texturing in quake3.  Some code was left over from the dirty/clean list
      The previous code would emit a full set of state during the first EmitState on
      Refactor the linux-dri config files so that linux-dri-x86 inherits from
      Add sis to the list of DRI drivers, since it had already been converted to the
      Fix SiS AGP vertex dispatch by not trying to emit 0 vertices, which would hang
      Correct a couple of comments.
      Triangle stipple is a fallback, since we don't have code for the stippling.
      Can't bail in sisDDLogicOpCode when logicop is off, because it's called with
      Bump driver_date for today's fixes.
      OK, one more time.  Simplify the state-backup system by just storing the full
      Bugzilla #1058:  Fix some potential 64bit pointer issues by storing differences
      Add an option for vertices emitted to be swapped CPU_TO_LE32, to be used by the
      Convert Rage 128 over to t_vertex.c.  While it's slightly slower (10% in ipers)
      If an application cleared before any state had been emitted, that clear would
      Remove empty r128_vb.h.
      Add Roland Scheidegger's S3TC patch.  This patch does not implement the
      Add support for NV_blend_square, and print errors if an unsupported blend
      Don't compile WriteMonoDepthSpan if HAVE_HW_DEPTH_SPANS is defined -- it won't
      Add fallback debugging (R128_DEBUG=fall) output, and set DO_DEBUG=1 by default
      Use the right FALLBACK macro for projtex so that projective textures actually
      This statechange is vtx state, not tcl.
      Always turn on the TAM_DEBUG3=0x6 workaround on real r200s.  It appears that
      Add code to support projective texturing and fix mixed enabling of texture
      Use the generic __amd64__, instead of __AMD64__ which has to be defined by hand.
      Add FreeBSD DRI build targets.  Doesn't build at the moment, due to the same
      Don't forget to -L/usr/local/lib to grab expat.
      Change another usage of __AMD64__ to the standard __amd64__, and remove

Felix Kuehling (99):
      Fix TCL_LIGHT_MODEL_CTL setting in radeonColorMaterial.
      Reenable the vtxfmt code paths in the radeon and r200 drivers.
      New, less expensive workaround for state-change lockups.
      Added two new options: def_max_anisotropy and no_neg_lod_bias. The latter one
      Added def_max_anisotropy and no_neg_lod_bias options to radeon and r200.
      Use hand-made number parsing functions as the libc routines are locale-dependent. Thanks to Pavel harry_x Palat for pointing that out.
      Fixed a few typos in the german translation.
      Parse radeon, not r200 configuration in radeonCreateContext.
      Make the drivers using the common texmem code work with NewTextureObject
      Include context.h in radeon_state.c.
      Imported the Savage DRI driver from the savage-2-0-0-branch of DRI CVS
      Call _tnl_allow_pixel/vertex_fog in savageCreateContext. (Brian Paul)
      Converted vertex setup to use t_vertex.[ch].
      Forgot to remove savagevb.h in last commit.
      Adapt the VERT_SET/COPY/SAVE/RESTORE_RGBA/SPEC macros to copy with all savage vertex formats.
      Set ctx->Const.MaxTexture[Image/Coord]Units correctly. Fixes corruption in q3.
      Reorganized hardware state data structures and state emission. Don't track
      Corrected the maximum vertex size argument of _tnl_init_vertices.
      Fixed some small state tracking issues while hunting a different problem.
      Cope with float colors in VERT_SET_RGBA and VERT_SET_SPEC macros.
      Clearing rectangle was one too big in each direction.
      Small texture management fixes. Should get rid of assertions some people were getting in savageBindTexture.
      Fixed two small problems in the vertex setup.
      Rewrote tiled texture upload. Small mipmap levels work correctly now.
      Fixed texture tiling format for 8bit texels on ProSavage and Twister.
      Swapout textures who's space is claimed by a new texture. Don't destroy.
      Cleaned up and slightly optimized the texture upload code.
      Cleaned up and fixed stencil fallback.
      Fixed MAXFIFO_S4. Removed WAIT_IDLE_EMPTY from savage_BCI_swap which resulted
      Fixed texture LOD bias.
      unsigned int -> uint32_t, unsigned char -> uint8_t where the size matters.
      Removed some useless definitions, artifacts from the Utah-GLX i810 driver.
      Improved MaxTextureLevel computation to take the nr. of tex units and both
      ctx->Const.MaxTextureLevels is not the highest mipmap level but the number
      Buffer vertices and emit them in batches. Still using conventional drawing
      Missing in last commit.
      Make sure hardware culling is disabled for unfilled primitives,
      More state management changes:
      Recalculate the viewport matrix in savageEmitDrawingRectangle.
      Enable drawing directly to the (now tiled) front buffer.
      Forgot to set SAVAGE_NEW_CULL in savageDDEnable_s3d.
      Define empty _tnl_x86_exec_vtxfmt_init if USE_X86_ASM is undefined. This
      Added german translation to option texture_units.
      WAIT_IDLE_EMPTY in savageSwapBuffers until real frame throttling is possible.
      Make GET_PROGRAM_NAME work with uCLibc using the __progname hack.
      Fixed hash table allocation to avoid an assertion failure due to a
      Converted the Savage driver to the new DRI interface and enabled compilation
      Fix emitting fog without secondary color and vice-versa.
      Fix strange white space that gcc didn't like.
      Fixed some missing uint8_t -> u_int8_t.
      Moved some code from savageCreateContext to savageInitDriver, where it
      Fixed initialization of some stencil-related S4 registers, which was
      Fixed GL_CLAM_TO_EDGE. It's not supported in hardware. Use GL_CLAMP
      Added configuration support to the Savage driver. Three options are
      Added a TNL pipeline stage that normalizes texture coordinates as a
      Made debugging output controllable via environment variable
      - Fake projective textures on a single texture unit. A fallback is only
      Enabled hardware rendering of 1D textures. No need for a software fallback.
      Simplified and optimized _savage_texnorm_stage.
      Use DRM headers from $(DRM_SOURCE_PATH)/shared-core instead of .../shared.
      Removed all direct hardware access (MMIO, BCI) from the Savage DRI
      Improved the performance of software fallbacks by not waiting for idle
      Added a boolean option for a floating-point depth buffer.
      Added support for floating point depth buffers on Savage4-based
      * Fixed handling of scissors
      Fixed: ptex drawing functions were overridden with standard drawing
      Added state debugging option (SAVAGE_DEBUG=state).
      Added missing break; to fix infinite recursion between choose_emit_func
      * Flush and wait in per-primitive fallback functions. Fixes flickering
      Fixes to cope with the lack of AGP textures. This is all that's needed
      Make sure DMA buffers are released in savageDestroyContext. Usually
      Use HW_READ_LOCK and HW_READ_UNLOCK in assembler-optimized span read
      Converted to use spantmp2.h.
      Added a fast path for emitting unclipped primitives directly to a
      Added support for ELTS to the _savage_render_stage. Requires at least
      * Added options for disabling the fast path (render stage) and vertex DMA
      Small cleanup:
      Fixed polygon offset with reversed viewport depth range.
      My last attempt to fix polygon offsets with the reversed viewport depth
      Fixed initialization of draw destination to front buffer on single
      Converted the Savage texture management to use Ian's common texmem code.
      * Fixed uploading of textures of certain sizes.
      Fixed some debug output that was unconditionally enabled.
      Fixed flushing of client-side vertex buffers.
      Don't try to kick textures from a NULL tex heap.
      Check for heap == NULL before driSetTextureSwapCounterLocation.
      Use all texture heaps in a fair way when textures need to be kicked in
      Made subtile upload a template. This speeds up tiled texture uploads by
      Added an option texture_heaps that allows selecting which texture heaps
      Fixed a typo in an assertion. (Don't understand how this ever compiled.)
      - Fixed scissor rectangle not moving with the window (Xorg bug #1731).
      Disable vertex DMA on SuperSavages.
      Fixed stride of incomplete tiles.
      Savage3D-based chips seem so use a constant tile stride of 2048 for
      Make glean's masked clear test pass.
      Offset LOD bias by about 0.3 to match software rendering more closely.
      Fall back properly when GL_COLOR_LOGIC_OP is enabled and != GL_COPY.
      With Savage DRM version 2.3.x or later use event counters for texture
      Don't forget to update a heap's timestamp when a texture image is

Gareth Hughes (104):
      Added GLU 1.3 tessellation (except winding rule code).
      Fixed triangle output, recovery process termination.
      Fixed file headers.  Tracking down macro bugs.
      Fixed definition of GLUtriangluatorObj for 1.3 tessellator.
      Fixed GLUtriangulatorObj again (spelling).
      Fixed legacy gluNextContour impact on gluTessEndContour.
      Win32 fix by Eero Pajarre.
      Added debugging output.
      Fixed vector macro calls, specifically COPY_3V params.
      Added debugging pragma message.
      Misc winding rule bug fixes.
      Winding rule updates.
      Added tessellation winding rule files.
      Added tessellation winding rule support.  Misc bug fixes.
      Fixed overflow bug in combineCallback.
      Updated debugging output.  I'm going to change it all to something
      Allow debugging output capture under Win32.  This seems really
      Correctly initialized GLUtesselator user data pointer.
      Renamed GLUtesselator user data pointer to avoid confusion.
      Updated memory allocation to allow new macros to be used.
      Fixed delete_all_contours memory deallocation.
      Added edge flag callback support.
      Removed CVS logs from files, as they were getting rather large.
      Added point_contour_test, taken from O'Rourke 1998.
      Changed ($MAKE) to $(MAKE) to allow real-clean to work correctly.
      Updated demo for new GLU 1.3 tessellation.  Added optimized rendering
      Major winding rule updates, especially stacked contour support.
      Added filename, line number output for debugging messages.
      Merging in work from 3.1/3.2 branch.  Tessellator is essentially fully
      Merged in fixes from 3.2 branch.  Fixed clipping bug, segfault in
      Merged 3.2 updates, namely combine callback for intersecting
      Removed '#if 0' from second pass.
      General cleanups.  Disable gl_katmai_project_vertices and
      Added SSE prefetching instructions.  General cleanups.
      Added SSE streaming store instructions, external symbol directives.
      Major audit of all Mesa's x86 assembly code.  This round is basically
      Major audit of all Mesa's x86 assembly code.  This round is basically
      Change animated apps to double-buffered visuals by default.
      Added trivial CVA test for testing HW driver fastpaths, good for testing
      Change transformation from ortho to perspective so we get the MGA elt
      - Fix tnl/t_context.h inclusion.
      Add basic timing information.
      Enable double-buffering, count not even multiple of three.
      Merge Mesa 3.4 test for operating system support for SSE.
      Allow operating system SSE support test to be overridden with an
      Add GL_POINTS as a primitive, including with CVA DrawElements.  Useful
      Implementation of GL_EXT_texture_env_dot3.
      Upgrade to (patched) version 7 of glext.h
      Some more work on interal debugging, timing routines for things that
      Force texture download every time to handle lazy texture downloads.
      Add missing copy in _mesa_noop_Color4fv().
      More generic vtxfmt template, useful for creating neutral tnl module as
      Add a couple of missing functions.
      Support for swappable tnl modules.
      Clean up install, restore for exec vtxfmts.
      Minor cleanups.
      Consistent copyright info (version number, date) across all files.
      Add missing header file include.
      Add missing header file.
      - Add -D_POSIX_SOURCE -D_POSIX_C_SOURCE=199309L to linux builds, we've
      Undo -D_POSIX_SOURCE change, fix the way the katmai support test is
      Nope, -ansi spoils that...
      Initial templates for immediate mode fastpaths, or custom tnl modules.
      Fix typo.
      - Port 3.4 texture utils, texture format work to 3.5 (including new
      Remove old code, fix a few comments.
      Fix header info.
      Fix error message.
      Make assertions take new GLchan formats into account.
      Remove #endif.
      Add missing header file include.
      More updates.  Radeon tnl module still disabled by default.
      Make COPY_FLOAT available to a wider audience.
      Revert back to new tessellator, check version string to be sure we're
      Make ESC quit.
      Fix compilation with GLU 1.1
      Add PolygonOffset driver callback.
      Fix texture format enum values (Brian Paul).
      Minor updates.
      Remove formats that no longer exist.
      Initial demo of GL_EXT_texture_filter_anisotropic extension.
      Remove compiler warnings about ASSERT_OUTSIDE_BEGIN_END() macro.
      Some more minor cleanups of macros.
      Fix ASSERT_OUTSIDE_SAVE_BEGIN_END() macro as well.
      - Minor cleanups of ctx->Driver.Current*Primitive usage.
      Updates required for DRI drivers on mesa-3-5-branch.
      Delete libGLU.so* from ../lib before installing it.
      More informative display of texture parameters.
      More texture format updates.  Drivers now need only plug an appropriate
      New type system for assembly code.  Asm files should now include
      - Restore texImage->IntFormat.
      Add missing GEN_SOURCES definitions.
      Make generation of matypes.h more transparent.
      Consolidation of asm code in 3.5
      Remove all traces of CULL_MASK_ACTIVE.
      Initial new Win32 build support.  More to come.
      More cleanups.
      Add new tests.
      Fix long-standing FP exception bug in _mesa_x86_cliptest_points4().
      Initial commit of cliptest work.  More to come shortly.
      Get rid of unfinished demo.
      Remove warnings about the difference between function pointers and
      Add missing items from .cvsignore files, tests/Makefile
      - Clean up state bitmask.

Holger Waechtler (3):
      added X86/vertex.S and X86/vertex_3dnow.S to ASM_SOURCES
      added prefetchw
      added CONCAT(x,y) macro in NASM/MASM section

Ian Romanick (235):
      Removed an accidental addition in the previous commit.
      Added support for NV_light_max_exponent.
      Removed fields from 'struct gl_extensions' for extensions that don't
      Removed all RCS / CVS tags (Id, Header, Date, etc.) from everything.
      Added support for GL_IBM_multimode_draw_arrays.
      Trivial changes to add support for GL_ARB_point_sprite, which is a
      Added support for EXT_texture_mirror_clamp and the single wrap mode
      Added most of the infrastructure required to support
      Fix to work with the C preprocessor in GCC 3.3.x.
      Removed redundant extension strings.  The Mesa extension mechanism
      Added some missing extension strings.
      Removed some dead code.  R128 does not (as far as I can tell) support
      Added support for EXT_texture_mirror_clamp.
      Remove dd_function_table::BlendFunc.  All drivers now use
      Silence a compiler warning DRI builds using gcc3.
      Make the R200 driver work with the NewTextureObject in Mesa.  Other
      Silence some compiler warnings.
      Corrected a comment about how the DOT3 hardware works in the R200.
      Modified error messages in _mesa_BlendFuncSeparateEXT to note that the
      Added support for EXT_blend_equation_separate / ATI_blend_equation_separate.
      Missed this commit as part of the addition of
      Fixed a cut-and-paste bug from the BlendFuncSeparate driver interface
      Refactor "class" texture environments to be implemented in terms of
      Added yet another version of gears.  This one uses fbconfigs, and is a
      Initial pass at adding support for the new DRI driver interfaces to
      Added a test case for Mesa bug #508473 that I wrote a long time ago.
      Enable use of new DRI interface in MGA driver.
      Fix a problem that caused the new interface code to not actually be
      Minor tweaks to get the fbconfig list returned by __driCreateNewScreen
      Refactored several variables out of XMesaVisual in favor of identical
      The available GLX version is now logged in addition to the client
      Create all the right modes for 16-bit screens.
      Fix a problem with 24-bit screens.  Modes with 0 stencil bits were
      Create all the right modes for 16-bit screens.
      Convert all uses of CARD32 and CARD8 to int32_t and int8_t.
      Convert int(8|32)_t to uint(8|32)_t, like should have been done the first time.
      Make sure drivers don't dereference a NULL pointer if
      Add #ifdef protection around the file to prevent DRI build breakage.
      Conditionally compile code for x86 specific TNL codegen.  This fixes problem
      Missed one with the last commit.
      Correct the comments about when the bindContext3 / unbindContext3
      Convert R200 driver to use the t_vertex interface.
      Remove include of texutil.h to get build working.
      Use the dispatch table directly instead of calling GL functions.
      Remove several files that are only used in the XFree86 / Xorg libGL.so.
      Added support for the 3rd texture coordinate for cubemaps and 3D
      Modified to use vertex arrays.  Vertex arrays and immediate mode calls
      Remove a debug printf that should not have been left in.
      Refactor "classic" texture environments to be implemented in terms of
      Added a config option to set the number of available texture units.  A
      Add support for the 3rd and 4th texture units.  The actual number of
      Enable use of new DRI interface in i830 driver.
      Uses either the GLX_SGI_make_current_read or GLX 1.3 interface,
      Warning clean-ups.
      Add support for GLX_SGI_make_current_read.
      Fix a possible segfault caused by the SGI_make_current_read changes.
      Added simple NV_blend_square test.
      Added simple EXT_blend_minmax test.
      Added a 4th square to the output.  Expanded the message written to the
      Fixed a bug that showed up in several multitexturing games (Quake3 and
      Added support for GL_EXT_blend_equation_separate and
      New scripts for processing the XML version of APIspec.  Mail is being
      Re-ordered categories.  Added extension number information.
      Added some in-code documentation.  Modifed glParameter to be a
      Added all of the missing ARB extensions.  Added all the enums and
      s/sop/rop/ in a couple places to fix some bugs.  Added some parameter
      Generates assembly dispatch stubs for SPARC.  Generates different output
      Fixed a bug that caused every dispatch function to call glNewList.
      Modify the generate assembly code to not use global registers %g2,
      Added a couple fog coordinate related register defines.
      Test the overhead of just calling a GL function.  The Python script is
      Minor code simplification.
      Convert all calls using _glapi_Dispatch to use the new GL_CALL macro.
      Modify glprocs.h to have two tables instead of one.  The first table
      Build the fogcoord demo.
      Move dri_util.[ch] and glcontextmodes.[ch] from dri_client to common.
      Define PFNGLXGETMSCRATEOMLPROC so that DRI will build without including
      Fix a couple typos that prevent building with DRI_NEW_INTERFACE_ONLY.
      Replace all occurances of XF86DRIClipRect (and related typedefs) with
      Replace occurances of GLXDrawable and Drawable with __DRIid.
      Fix a couple issues related to GetDrawableInfo.  In the new DRI
      Convert 'Display *' to '__DRInativeDisplay *'.  Only portions of the
      Fix dumb mistake from a previous commit.  __driCreateScreen is now
      Replace drmHandle, drmContext, drmDrawable, drmMagic and related types with
      XF86DRIDestroyContext, XF86DRICreateDrawable, and
      DRI_NEW_INTERFACE_ONLY is now the only support way to build in the
      Re-import from DRI tree.
      Replace use of XID with __DRIid.
      Minor changes to make it work better outside X.  "Success" is replaced
      Replace Bool with GLboolean.
      Make sure USE_NEW_INTERFACE is defined if DRI_NEW_INTERFACE_ONLY is
      Require libGL API version 20040602, which actually exports
      driCheckDriDdxDrmVersion uses a function that is not available to
      Add a new function, driFillInModes, that is to be used by drivers to
      Remove fill_in_modes and use driFillInModes instead.
      Enable use of new DRI interface in Rage128 driver.
      Enable GL_EXT_blend_subtract and GL_NV_blend_square.
      Replace 'Display *' with '__DRInativeDisplay *'.
      Start to "modernize" the driver.
      Enable use of new DRI interface in Unichrome driver.
      Bool may not be available when DRI_NEW_INTERFACE_ONLY is defined.
      Remove drivers that have not been converted to the new interface from
      Remove the rest of the need for glxclient.h and remove it from the tree.
      The required DDX and DRI version numbers were switched.  This went unnoticed
      Fix up some remaining include issues.  Replace Xmalloc / Xfree with
      Fix a couple typeos in the tables for driFillInModes.
      Make the test check for core GL version 1.3 in addition to testing for
      Don't include glapi_*.S files in DRI / solo builds.
      Make sure mklib sees the definition of CC and CXX.  Make mklib respect
      Re-enable SPARC assembly on sunos5-gcc platform.
      First phase of TLS work.  At this point SPARC assembly dispatch is
      Make sure THREADS is set no matter which threading model is selected.
      Fix a potential race condition.  Fix a couple of places where
      Another "if it ain't broke, don't fix it" moment.  Revert some more
      Fix typos in comments.
      Fixed an assembler warning / error (depending on the version of GAS used).
      Minor extension string tweaks.  Added support for vertex program extensions.
      Merge in all the i830 functional differences from the old i830 driver.
      Added GL_ARB_point_parameters support for i830.
      Add GL_ARB_texture_cube_map support for i830.  Most of the code was
      Factored out code for setting texture coordinate wrap bits.  Added support
      Replace the numerous switch-statements to derive GL_COMBINE post-scale
      Cut out all of the old texture environment code and rewrote it from
      The alpha post-scale and the RGB post-scale were mistakenly reversed.
      Set MaxTextureImageUnits and MaxTextureCoordUnits to match MaxTextureUnits.
      Add some Savage3D stencil documentation based on discussions during the
      Also export fbconfigs with 0/0 depth/stencil modes.  This fixes "driver
      Use the driFillInModes utility function.
      Convert tdfxDDInitExtensions to use driInitExtensions.
      Prevent Y-offset from exceeding valid range in texture upload code.  This
      Add simple test program for Mesa bug #1028405.
      Reject unsupported texture formats passed to glCompressedTexImage?D.  This
      Massively cleaned up the code that calculates front/back/depth buffer
      Eliminate the funky SSE exception test from DRI builds.  It's not need
      Refactor the code in driGetRendererString a bit to make it easier to
      Initial support for PowerPC specific code in Mesa and DRI drivers.  DRI
      Make the GL_RENDERER string include the actual chipset information
      Added some documentation that I managed to figure out through
      Add support for optimized versions of the code underlying ReadPixels
      Convert MGA driver to use new spantmp2.h interface.
      Fixed a few places that should have been using GET_SRC_PTR.
      Only build the MMX/SSE/SSE2 versions for ReadRGBASpan when the pixel format
      Add support for OES_read_format.  As soon as glext.h is updated with the
      Fix hangs on big-endian (e.g., PowerPC) hardware.
      Add ARGB modes to support big-endian systems.
      Fix t_vertex byte-ordering issues for PowerPC.  This was tested with gears
      Fix compile errors when DEBUG is defined.
      Small optimization for big-endian (e.g., PowerPC) systems.
      Big-endian fixes for R200 sw TCL path.
      As per suggestions my Michel Daenzer, improve the fix for bugzilla #1513.
      Added MMX optimized version of the RGB565 ReadRGBASpan routine.
      Add EXT_stencil_wrap test that doesn't depend on stencil read-back
      Set the "ignore" flag for any functions that the open-source libGL doesn't
      Added information about numerous enums to aid the development of automatic
      Added support for several additional enum-only extensions:
      PrioritizeTextures, GetClipPlane, and WindowPos3fMESA don't need to be
      Since we know the problem with HP_occlusion_test is / was on the server-side,
      Missed GL_OES_read_format_bit in last commit.
      Add trivial support ARB_texture_rectangle and OES_read_format.
      Added enums for GL_OES_read_format.
      Added the ability to get the size of a parameter as a string.  Changed the
      Make the transition to script-genereated GLX code easier.
      Make the transition to script-genereated GLX code easier.
      Make the transition to script-genereated GLX code easier.
      Make the transition to script-genereated GLX code easier.
      Make the transition to script-genereated GLX code easier.
      Python script to generate various bits of client-side GLX protocol code.
      Make the transition to script-genereated GLX code easier.
      Make the transition to script-genereated GLX code easier.
      Use the indirect_init.c generate by using src/mesa/glapi/glX_proto_send.py
      Slightly tweak the format of the size prototypes.
      Use indirect_size.c and indirect_size.h generated by using
      Refactor ExtractDir and OpenDriver to be a bit more clear.
      Prevent possible divide-by-zero error.
      Added a couple missing always_array annotations and a couple trival
      Corrected a problem with the sizes of the MAP[12]_* enums.
      Updated the schema.  Added comments explaining some of the non-obvious
      Fix a minor bug in glXEnumFunction::PrintUsingTable.  Add some comments.
      Move common GLX code to glX_XML.py.  This will make adding glX_proto_recv.py
      Trivial shell script to search the API definition file and print out the
      Added many (hopefully all) missing 'output="true"' attributes.  Added
      No GLX protocol for GL_ATI_fragment_shader.
      Use GLX protocol code generated by glX_proto_send.py.
      Added some comments and fixed typeos.  Slightly refactored the way
      Used GCC's __builtin_expect when available.  Change the way code is
      Fixed two problems with the handling of GLX protocol replies.  The logic
      The opcode and command length fields of RenderLarge commands were
      Revert some accidental cross-merge changes.  Remove i915GetString.  Migrate
      The generic_*_byte functions did not rount the command size to a
      Pixel oriented render functions are now generated by the
      Refactor the code that converts a transpose-matrix enum to a
      Track the GLX protocol names for GL types.
      Script to generate rough GLX protocol documentation (as might be seen in
      Use GNU AS label aliasing, if available.  This must be enabled by adding
      Put quotes around the CC and CXX variables passed to mklib.  This make
      Added a couple missing proxy types.
      Make sure that functions added to gl_API.xml that do not have any GLX
      Since only one of the places that used glXFunction::command_payload_length
      Remove glEnum::startElement, but refactor out some of the common code from
      Correct some of the dependencies and clean things up a bit.
      Move all the code for generating "size" related functions to its own
      Correctly set the name of the generating script.
      Add a "count" attribute to "enums" elements to set the default count
      Stupid text editor...convert wayward tabs to spaces.
      Add a glFunctionIterator class to iterate over the functions stored in a
      remove a bunch of temporary #defines, now that glext.h was updated
      The 'mode' setting of a function within an 'enum' element is now used.
      Add numerous 'get'-type functions to most of the enums supported by the
      Mark a couple more function parameters as counters.
      Add an option to emit extra debug code in the generated protocol functions.
      Slightly modify the meaning of the 'handcode' attribute in a 'glx' element.
      Fix some get / set errors with COLOR_TABLE_*.  SCALE and BIAS can be
      Properly annotate variable-length output parameters.  Make sure that the
      Refactor the code to emit multiple-inclusion protection to
      Add GlxProto::createEnumFunction and add a 'context' parameter to the
      Small refactor.  Move several utility functions from glX_XML.py to
      Small refactor.  Add glXFunctionIterator, which derrives from
      Added a bunch of extensions that were previously supported but weren't
      Enable client-side GLX support for texture compression extensions.
      Fix some mixed spaces / tabs issues in generated code.  Commit generated
      Add the ability to include other XML files by using a
      Definitions for EXT_framebuffer_object.  Just <include/> it in gl_API.xml
      Correct a couple problems / omissions in the DTD.
      Generate GLX protocol for pixel single commands.
      Simple test for ARB_texture_env_crossbar functionality.  It is difficult
      Simple test for GL_MESA_pack_invert.
      Added __glExtensionBiIsEnabled and __GLXcontext::gl_extension_bits.  This
      Missed the Makefile with the previous commit.
      Corrected the value for X_GLrop_SecondaryColor3svEXT.
      Put quotes around $(CC) and $(CXX) so that 'CC=cache gcc' will work.
      Missed the commit of glxext.c yesterday.
      Convert the static functions read_pixel_reply, read_reply,
      Add GLX protocol support for ARB_fragement_program,
      Accidentally committed wrong version with the previous commit set.
      Fairly significant changes to enums.c and the way it is generated.  enums.c

Jerome Glisse (4):
      Modified hw state for big endian computers.
      Added PCI id (0x4152).
      Added PCI id (0x4152).
      Added texture format table for big endian (this is a tempory hack).

Jon Smirl (37):
      Renaming Makefile to Makefile.X11, added depend support, adjusted paths
      Add file miniglx.h
      Added missing includes for standard header files
      Rename Makefile to Makefile.X11, add depend support
      update fb makefile for new directory structure, add depends
      Remove files not used in current MGA driver
      Add missing assembly file from radeon build
      Fix compile problems with `glMultiModeDrawArraysIBM'
      Add cast to remove compiler warning
      Minor changes to make linux-solo build again
      Define SOLO so that check is possible for SOLO in project makefiles
      Fix Makefiles to copy lib to $(TOP)/lib if missing
      Update DRI drivers to current DRI CVS and make them work.
      Clean out some unused files
      Fix for missing GLAPIENTRYP definition
      Adjustments to make everything use IOCTL/sarea defines in DRM instead
      These files are redundant now. Definitions come from DRM files.
      Make R128 driver use sarea/defines from DRM. removed r128_common.h
      Adjust mga drivers to remove redundant h file for sarea and IOCTLs
      Adjust includes to help DRI build
      Add DRM_SOURCE_PATH to make system.
      Fix solo build, add missing #ifdef NEW_INTERFACE
      Removed need for sarea.h, various touch ups to get rid of type mismatches.
      kill __driRegisterExtensions(), none of the functions did anything
      Open/Close FullScreen die. unichrome and savage implemented, code is ifdef'd out
      Redo of removing __driRegisterExtensions(). I accidentally undid the previous check in.
      Mesa-solo builds with these changes. There are still more fixups needed to
      This should make most linux-solo drivers work again. Mainly a fix
      Patch removes _SOLO definition needed for mesa-solo. mesa-solo
      Convert miniglx to use Ian's fancy new driver config code.
      Convert miniglx to use xf86drm from dri/drm/libdrm instead of src/mesa/drivers/dri/dri_client
      Fix mesa-solo config to define _GNU_SOURCE so that program_invocation_short_name
      Fix unichrome solo build to not need X11 headers. Minor change to
      Remove DRM_PAGE_SIZE from fb and i810 drivers since it was being used.
      Make solo radeon/r128 drivers track fix for DRI bug 849
      Miniglx clients now authorize with server DRM
      Fix from Rogelio Serrano for miniglx double buffering

Jon Taylor (30):
      *** empty log message ***
      *** empty log message ***
      *** empty log message ***
      *** empty log message ***
      *** empty log message ***
      *** empty log message ***
      *** empty log message ***
      Updated to reflect GGIMesa 3.1 changes
      Misc small cleanups
      *** empty log message ***
      Fixed double buffering bug when running in X
      Re-added missing GGIMesa context bufsize structure member
      Disable genkgi sublib handling until we have some sublibs to handle
      * Fixups to the GGIMesa code to reflect the recent DD API changes
      Minor fixups
      Lots of small updates to GGIMesa, tracking LibGGI internals changes
      Lots of misc. bugfixes to bring GGIMesa in sync with the latest Mesa sources.
      * Updated the autoconf/automake/libtool build system
      Small compile fix
      * Auto* build fixes
      Added GGI-style debugging harness to GGIGLUT.
      * More work on the auto* build system
      More auto* build fixes
      swrast setup fixes
      More auto* build system fixes
      More auto* build system updates
      Added a new configure.in flag, --enable-glut-build.
      Fixed GGIMesa build problems.
      Added missing #include "mmath.h"
      Misc build fixes

Jose Fonseca (12):
      Remaining patch for the 255x255=255 identity correction be
      Removed code that was proven to have faster alternatives to reduce complexity and facilitate reusability.
      Definition of several utility macros for self-contained MMX operations such as scaling and lerping.
      MMX modulate belding function added.
      MMX add blending function added.
      Added min and max blend functions.
      Re-scheduling of the instructions according to the MMX pairing rules. This is dificult to quantify but the improvemnet reaches about of 0.7% in some cases in a Pentium III. In older Pentiums the improvements could be quite substancial due less capable integer pipelines.
      Merge the Doxygen configuration files from the embedded branch.
      Seperate the Doxygen configuration files (for full Mesa only) in a common part and an individual module part.
      Change the hyperlinks names in the main page to match the Doxygen generated ones.
      Update the doxygen configuration file.
      Use CSS in custom headers.

Josh Vanderhoof (7):
      features weren
      no underscore prefix for freebsd >= 3 (Simon Walton)
      swap operand order for Intel style REGOFF
      fixed SSE bugs
      x86 assembly dispatch
      no brackets for nasm indirect jumps
      prevent spaces in fp register names

Jouk Jansen (102):
      Updated
      update
      Update after disk crash
      Update after disk crash
      This time the correct file
      Added vertices.c to compile list
      strcmp needs the string.h
      mesa_wgl.h should only be included when _WIN32 is defined
      Added tess_winding.c for compilation
      misc -> glmisc
      #include <assert.h>
      #include <stdlib.h>
      Repaired definition of GL_CALLOC_STRUC
      tess_macro.h included
      version 3.3 update
      Workaround for the VMS 32 character maximum limit.
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      ----------------------------------------------------------------------
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Modified Files:
      Modified Files:
      Modified Files:
      Committing in .
      Modified Files:
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      VMS compile update
      Committing in .
      Committing in .
      Committing in .
      Modified Files:
      Modified Files:
      Committing in .
      Committing in .
      Modified Files:
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .
      Committing in .

Karl Rasche (22):
      Add extra cases and enums for _mesa_fetch_state(), and set the length of
      Initial checking of new ARB_frag/vertex program parser
      Initial checkin of new ARB_frag/vertex program parser
      Initial checkin of new ARB frag/vertex program parser
      Caught missing parse_color_type() [Jacob Jansen]
      Porting NV_vertex_program tests to ARB_vertex_program
      Porting NV_vertex_program demos to ARB_vertex_program
      - Create a dummy program to keep from segfaulting when parsing
      - parse_alias() Was totally borked
      Moving NV_fragment_program tests to ARB_fragment_program
      added support for generic vertex attributes (ARB_vp) and their error checking
      - New version of the syntax rules from Michal Krol
      Fixed compiler warnings
      Remove things having to do with ARB_matrix_palette/ARB_vertex_blend
      s/inverse/invtrans.. whoopsie
      A quick little test to show disapeearing specular color between vertex and fragment program execution
      - handle IsPositionInvariant
      - New grammar (0.5) from Michal Krol, merges many opcodes and adds range testing
      added parsing exercises for ARL and relative offsets in ARB_vp
      Arg. I can't code.. Test that we don't recongnize '0' as having a leading zero..
      - Test instruction count
      Problem computing specular value in OPCODE_LIT - pulled in value from

Karl Schultz (182):
      compiler warnings
      compiler warnings
      remove carriage return chars
      compiler warnings
      clean up makefiles some more for Windows
      more compiler warning fixes
      silence compiler warnings (last batch for src)
      Change UnProject4 prototype to use nearVal and farVal instead of near
      compiler warnings
      Don't use "errno" as a parameter name - it is also a function and that
      add mangling for new entry points
      add the same conditional mangling stuff that gl.h has.
      Add GLAPIENTRY to typedef for callback functions (used by tessellator).
      Make osmesa a DLL instead of a static lib (Windows).  This is more
      use OSMESA_* enums for OSMesaCreateContext instead of GL_*.
      disable some more code so that it compiles quietly on Windows.
      Fix up Windows makefiles so that the debug build leaves debugger info
      remove extra CR chars
      Updated Windows driver for Mesa 4.0.
      Need to define MESA_DEBUG in debug builds.  Otherwise, NDEBUG gets
      Build a few demos that now work with Windows.
      Allow to compile in Windows.
      Fix the clear() function
      Fix comments.
      Updated for Mesa 4.0
      Copy the readtex.[ch] files from ../util
      Change name of glut lib from glut32 to MesaGlut.
      Updates for improved DirectDraw support (Daniel Slater)
      avoid using near and far as variable names since they are reserved words
      Fix compilation errors and warnings for NURBS support. (Robert Bergkvist)
      Update notes on si-glu NURBS.
      update to version 4.0
      main trunk is now 4.1
      malloc() prototype is in stdlib.h for darwin
      Add darwin target
      Corrected bad line breaks in macro definitions within code compiled
      Added missing entry points
      Remove const from save_TexParameteri to match prototype for glTexParameteri.
      Remove rules that were put into place to deal with compiling files that
      Fix up targets for
      Cleanup for IRIX 6 platforms.  Build src-glu instead of si-glu due
      Add .cc as a valid suffix.
      Use $(MAKE) on the irix targets, instead of hard-coding pmake,
      Define finite as _finite for Windows.
      Remove trailing CR's.   No logical changes.
      Fixed pixel color component problem and clear code for 24-bit Windows
      Remove phantom tab on a blank line to allow compile on IRIX.
      Correct typo (vpstate.c) so that makefile works.
      Removed missing files and added new ones (Robert Bergkvist)
      Demo updates for Windows (Robert Bergkvist)
      Samples working for Windows (Robert Bergkvist)
      get tests working on windows (Robert Bergkvist)
      Fixes for 24-bit Windows devices.  The old code was trying to do 3-byte
      fix compiler warning (windows)
      Add _glapi_get_context to export for osmesa.
      casts to silence compiler warnings
      export more symbols for osmesa (win32)
      Added -perf option to measure frame rate.  This is useful for measuring
      Update filelist to reflect two removed source files.
      Provide a reasonable replacement for __FUNCTION__ when using non GNU C
      Fix up alpha buffer handling for Windows.
      Apply alpha buffer control fix to accum buffer too.
      export new symbols for osmesa lib
      Catch up windows driver with recent context mods
      Coerce vsnprintf to _vsnprintf for windows
      Add note about needing to run VCVARS32.BAT when compiling.
      Don't call _mesa_initialize_context because we are using
      Mark destroyed context as not current.  (Frank Warmerdam)
      Fix for bug 618459.  Mods to allow compilation on VC++ 5.0 (Michael Krause)
      remove const storage class specifier for the decl of a var that isn't
      include procs for compressed textures
      Change for Windows only:  Make demos and examples build again.
      Put back the include of windows.h so we are back to version 17 of the file.
      Add casts to quiet compiler warnings.
      Add casts to quiet compiler warnings.
      Quiet compiler warnings.
      export new functions.
      Add casts to quiet compiler warnings.
      Add instructions to unzip both zip files into the same directory.
      Tweaks for hpux.  The hpux10 and hpux10-sl targets now build pretty
      Add _mesa_enable_1_4_extensions and _mesa_problem
      Fix SourceForge bug 664908.  wglDescribePixelFormat should return the
      Check only LEFT bits for clearing alpha buffer. (Evgeny Kotsuba)
      Fix function definitions for Windows.
      fix compilation warnings on Windows
      fix compilation for windows
      Windows does not define HUGE_VAL, so use the IEEE hex constants.
      Remove obsolete file.
      Removing from this directory.  Will be added to new gdi driver directory.
      Visual Studio Workspace file for building Mesa.  This workspace contains projects for building core mesa, backend drivers (e.g., gdi, dx7, ...), GLU, and GLUT.  The project files are placed in the appropriate places in the Mesa tree.
      Project file to build core mesa - creates static LIB that is later linked to make a DLL.
      Project file to build swrast - creates static LIB that is later linked to make a DLL.
      Project file to build tnl - creates static LIB that is later linked to make a DLL.
      Project file to build gdi driver and OPENGL32.DLL.
      Source code for GDI driver.
      Project file to build GLU32.DLL
      Project file to build GLUT32.DLL
      Workspace to build Mesa demos and sample programs.
      Project to build the gears demo program.
      linker exports defs
      fix location of mesa.def
      Microsoft Visual Studio 6 Workspace file for building Mesa and related libraries.  (Moved from src/mesa/drivers/windows to make it easier to find.)  The project files belonging to the workspace are located elsewhere in the tree, closer to the modules they each build.
      moved to top of tree.
      Project file for osmesa
      add symbols needed by osmesa.
      Add osmesa project
      project file for osdemo
      add osdemo project
      cleanup - add comments and sort names.
      add a dependency so that osmesa builds
      Project now builds the C++ sources (with great difficulty).  Add module definition file (glu.def) to project.
      Add documentation.
      Compiler/Linker "response" file containing compiler options and list of files to compile/link for the C++ files in the glu library.  It is easier to edit and maintain this infomation in flat files instead of trying to enter it into the VC project menus.  The C++ files are compiled as a "pre-link" step.
      minor option tweak
      VC 6 Project File
      Reorganized projects so there is one project for each Mesa source subdir, and put the actual project files in each subdir.
      Add the C++ and response files to the project so that they show up in the project file list.
      Moved.
      add arbfragparse.c to project.
      Silence compiler warnings about implicit casts or conversions by supplying explicit casts and/or tweaking constant and variable definitions.
      Generate browse info for Debug version.
      Add casts to prevent signed/unsigned compare compiler warnings.
      Add casts to prevent double->float conversion compiler warnings.
      Add symbols to export private entry points to osmesa.
      Copy the DLL files to the lib directory instead of libexec.
      Update with current set of gl* entry points.
      Remove first attempt at using visual studio projects to compile demos.
      New visual studio workspace and projects for building demos.
      Now obsolete.
      Updated for new Windows build system.
      Make binary - even though this is a text file, common practice is to store MS studio files as binary
      Make binary - even though this is a text file, common practice is to store MS studio files as binary
      Make binary - even though this is a text file, common practice is to store MS studio files as binary
      Add include for context.h to define NEED_SECONDARY_COLOR.
      New visual studio project fiile for building demos.
      Add remaining demos.
      Create 'all' project to make it easy to build and clean all the programs in the demos directory.
      Change projection transform so that the top row of teapots are now visible.
      Workspace file for redbook programs
      MS dev studio project file for teapots demo.
      Add project file for new spriteblast demo.
      Add project file for new spriteblast demo. (Update dependency)
      Add notes about building with the mangled namespace on Windows.
      Add missing entry points.  (Warren Cashen)
      Add missing files for building Windows distribution.
      Fix error in last check-in.
      Define GLAPIENTRYP properly so that GLU compiles on Windows.
      Add missing GLAPIENTRY
      Add missing GLAPIENTRY
      update files
      Change
      Windows needs to #define GL_GLEXT_PROTOTYPES
      remove ss_vb.c from project
      add t_vertex.c to project
      add casts to silence compiler warnings
      init an uninitialized variable.  This doesn't fix a latent bug because
      add cast to quiet compiler warning
      change type of loop index var to remove compiler warning.
      remove unused variable
      silence compiler warnings
      add new entrypoints
      Replace with version generated from new python script.
      Generate mesa.def file for Windows.
      add casts to quiet compiler warnings
      Add new driverfuncs.
      Avoid setting NOWIN98 compiler optimization with VC 7.
      Fix compiler options regarding exception handing to resolve warnings.
      Add GLAPIENTRY function decorations for correct operation on Windows.
      Minor tweaks to deal with vsnprintf and __FUNCTION__ on WIN32
      New Visual Studio project file for this directory.
      Update Visual Studio Project file for src tree updates.
      add shader.dsp and descrip.mms to tarball list for relatively new shader dir.
      rearrange a couple of lines of code to avoid compilation error in VC 6,
      Fix bug in glDrawPixels when writing color indexed images on color indexed
      Enable builds on AIX 64-bit
      fix problems in previous version
      HP shared lib config
      Add HP shared lib configs.
      Remove glut project from the Mesa workspace.  This allows users to build
      Add the glut project to the demos workspace.
      minor updates and improvements
      *** empty log message ***

Keith Whitwell (486):
      various pipeline bugs
      new files to support vertex-based geometry
      more fixes on the fast path
      removed bogus call
      fixed GL_FRONT_AND_BACK cull bug
      fix for GL_COMPILE_AND_EXECUTE bug
      Fog, displaylist and zoom operations
      more bugfixes
      fix for mga depthbuffer resize
      fixes for the mga driver
      new DD_Z_NEVER flag
      bugix from raul for fxddbitmap
      Do pop for GL_VIEWPORT_BIT via. the api
      Fix for write-rgba-span and others
      Large patch:
      RasterMask was set incorrectly
      Cleaning up
      Bugfix for ctx->backface_sign calculation
      added mesa.conf
      more hooks for mga driver, including an immediate fastpath
      Fixed includes & added a few hooks for the DRI.
      fixed voodoo initialization
      Changes to reduce the memory footprint of display lists
      Fix for glerror on compilation of list containing gldrawelements calls
      fix for (some of?) the vertex array problems
      second try at drawarrays fix
      more fixes for glDrawArrays
      use glClientActiveTexture in interleaved setup
      fix for colormaterial
      fixes for andree's drawarrays problems
      another typo
      Fog coordinate stage which drivers may use to replace standard fogging
      add callbacks for stipple
      fix for tristrip bug and unfilled quad/polygon bug
      Added stipple mode
      New clipping to avoid division by zero bug.
      Use smooth shaded triangles always.  Fix SoF bug.
      Re-enable the keyboard functionality.
      fix GL_BACK color material bug
      Performance fix for unfilled triangles on mga and i810 hardware.
      test
      test
      test
      test
      test
      test
      test
      test
      test
      test
      Fix for miscolored rects in display lists.
      Fix for gloss and sgl eval bugs.
      fix for eval regression
      Implement EXT_fog_coord and EXT_secondary_color.
      Do per-pixel fog when GL_FOG_HINT is GL_NICEST.
      Replace the flags Mesa was using for ctx->NewState with a new set
      test cvs commit logging
      Removed references to gcc's __FUNCTION__ macro.
      Rearrange VERT_* flags to support 4 texture units.
      Remove dead code in FX driver.
      Moved software rasterizer functionality to new directory.
      Moved the software rasterizer to a new directory.
      Lots of changes:
      A new module to provide RasterSetup and advanced triangle/line/point
      Reorganized software rasterizer as a module which manages its own state,
      Use the new software rasterizer.  Reworked optimized line,tri,point
      Reworked to use the new software rasterizer.  Optimized line/tri functions
      - Changes for new software rasterizer modules
      fixed very obvious fog bug
      Cleanup of derived state calculation prior to seperating software T&L
      Move the transform and lighting code to two new directories
      fix lvalue cast on ctx->swtnl_im
      Support for swappable t&l modules, including an example one in the FX
      New files...
      Fixed a couple of bugs that crept into last commit
      Major rework of tnl module
      fix xscreensaver cores
      The array cache.
      Fixes for compiling assembly (disable unused 'masked' versions)
      fix sproingies bug
      Fix evalpoints bug (samples/nurb)
      fix demos/fire, enable lazy vertex flushing
      try and force a log message
      try and force a log message
      try and force a log message
      try and force a log message
      try and force a log message
      try and force a log message
      try and force a log message
      Add render stage for unclipped vb's to fx driver.
      Remove 'pv' parameter from Line/Tri/Point funcs.  The provoking vertex
      various compilation/warning fixes
      Moved glcore.h to a location prefixed by 'GL/internal', for compatibility
      Modify X11 driver and fake glx to store a GLcontext
      Bring the FX driver into line with core mesa changes.
      Add a 'RenderPrimitive' callback to t_vb_render.c.  Helps out drivers
      revert XMesaCreateContext changes
      Add call to Driver.RenderPrimitive()
      Still need 'DD_TRI_CULL_FRONT_BACK'
      Fixed 'IRound' to 'IROUND' in mmath.h
      Resync with new XFree version of this file
      Fix crash in book/stencil.
      Use correct pv in swrast lines routines
      fix conform dlist test
      Fixed conform feedback and drawelements tests.
      Fix several conformance problems.  Hack solution to line stipple problem.
      Fixes for performance bug on compiled array element paths.
      Removed knowledge of swrast Clear/Bitmap/Accum/Draw/Read/CopyPixels
      Update implementations of Driver.Clear().
      Fast no-copy drawarrays for large tristrips
      rework to correctly respect _ac_import_range()
      Use a lookup table to compute exponents in tnl fogging code.  Slightly
      Correctly calculate size for disabled texcoord arrays.
      Fix propogation of material values in VB's that don't reach the lighting
      Fixed conform problems with recent material tracking change.
      Allow swrast to cope (fairly) cleanly with GL_SEPERATE_SPECULAR when
      Fix state bitmasks for invalidating line,tri,point functions.  (won't fix
      Added GLvector4chan type, removed lots of CHAN_TYPE ifdefs.
      Template work
      Initialize Line._Width
      Templates are in working order.
      Bugfixes for vertex format, templates
      Uncomment some defaults.
      Add missing cases for viewport transform.  Remove INVALIDATE_STORED_VERTICS
      Fix typo
      Changes for ffb
      DO_POINT renamed DO_POINTS
      Revert earlier changes protecting against null VB->TexCoordPtr[x].  Do
      Remove dead #define.
      Fix cut&paste bug exposed by clipbug.c demo.
      Fix lineloops.
      Split driver struct into swrast/tnl/core components.
      New files
      Removed DD_STENCIL, DD_FEEDBACK, DD_SELECT.
      Remove ENABLE_* flags, ctx->_Enabled.
      Fix calculation of fog coordinate in translate_vertex().
      Perform fixup on material values
      Fix striding of color material inputs.  (Fixes glean colorLitPerf)
      Remove values from VB->importable_data as those arrays are removed
      Add more options, including '-10','-100' and '-1000' to limit nr of vertices.
      no-copy drawarrays for remaining non-fan primitives
      first pass at eval fixes
      Support for floating point color representation in tnl module.
      add specialized interp for backcolor, edgeflags
      fix color copying in twosided quads
      Allow drivers to provide a dynamic CHECK_HW_DIVIDE, to turn the divide off
      More HW_DIVIDE checks.
      typo
      Restore optimization for cva glArrayElement operation.
      Lots more eval fixes
      fix from Jouk
      removed debug
      float color changes
      add disabled test to turn on vertex snapping.
      fix for moebius infinite loop
      fix eval/colormaterial interaction
      Don't scale basealpha by 255 now it's a float.
      Use correct PV when clipping.
      Add FLUSH_CURRENT() calls to fix ordering problems (incl. hierarchical
      fix possible segfault on destroy context
      add missing translate functions
      remove debug code
      Replace PipelineStart, PipelineFinish with RunPipeline.  Clean up
      Driver interface changes
      Set 'prefer_float_colors' true, now that tnl uses them natively.
      Clean up translation of array elements, copying of elts in pure-elt
      made render path selection more flexible
      fix a couple of small bugs
      Clean up _tnl_Begin/begin/hard_begin.
      increase blocksize, remove dead opcodes
      Fix for glean texgen test.
      typo
      fix for Jouk's glplanet bug
      Fix order of decomposition of quad.
      Comment out __FUNCTION__ usage.
      rerun vertex stage on _MESA_NEW_NEED_EYE_COORDS
      apply David Miller's quad/backcolor patch
      Working implementation of glArrayElt that decomposes to regular
      Add api_arrayelt.c
      New files to provide glEvalPoint, glEvalCoord implementations for
      fixed some GL_POINTS probs
      Adaptor code that lets tnl convert compiled (display list) cassettes back
      Add -f to rm in make clean to avoid error messages
      restore normal length optimization in dlists
      Rename some of the tnl->Driver.* functions to tnl->Driver.Render.*, to make it
      Lighting now emits colors as CHAN_TYPE, as it used to.  This will require
      Fix Alan Barnett's 'try10' display list bug.
      Remove _BaseAlpha, fix reflect lighting bug.
      Fix core in fixup_normal_lengths
      Fix copying problem (light spots) on evaluated surfaces.
      Fix SGL monoindexed bug (glDrawRangeElements in dlist)
      Fix SGL testGeoSets.exe polygon mode problem (wrong interpretation of
      Fix SGL testLights.exe bugs (interaction of copying and fixup in display lists)
      Fix VERT_RGBA special case.
      Removed optimizations (this code is now seldom used)
      Fix problem with colormaterial (bug #441859)
      Compute first partials before taking cross product to calculate normal.
      Undo last change - breaks other stuff.
      Reenable, but only in the size four case...
      Fix FX driver to the point where gears & morph3d work.
      remove some debug
      (very) Brief overview of tnl module
      very brief overview of swrast
      More of the same
      use correct vertex for flatshading
      forgotten diffuse alpha fix
      possible fix for cva ignore start index bug
      3dnow normalize_normals broken, disable
      Another locked drawarrys fix
      Possible fix for vtk tristrip bug
      use smaller buffer when decomposing large drawarrays, for cache goodness
      correctly reset input after dlist execution
      avoid side-effects in macro param
      simplify drawarrays paths
      fx another place where PRIM_PARITY is calculated
      guard ResetStipple calls
      further tweak to dlist change
      add inf and nan tests to swrast functions
      fix glut tess lockup
      fix primitive calculation in 1-line dlist immediates (molecules)
      fix for ubyte colors
      Fix clipping problem on mga
      Further help with dri libGL version skew
      add _compat_VertexAttrib4fNV
      More suport for t&l drivers
      Remove debug
      remove dead vertex assembly
      remove dead vertex assembly
      bring in changes from dri tcl branch
      Core of the future vertex engine.  Isn't built yet, nor will be for a while...
      Fix FETCH(argb1555) mask values
      another checkpoint of struct immediate replacement code
      FRONT_AND_BACK culling
      update colormaterial state on swtcl wakeup
      add GL_UNPACK_CLIENT_STORAGE_APPLE gets
      lineloop fix
      Fix projtex interp bug
      Add a note about window map behaviour and the autotuning of nr of loops which
      fix 3dnow misdetection on new p4's
      Flush IM if very small number of verts left -- prevent copy errors
      Don't crash if End received without a Begin...
      Add mechanism to destroy mutexes.  Important on OS's where mutex initialization
      Add missing break statements
      Fix typo
      Pass 4th element correctly in loopback_VertexAttrib4svNV
      Don't unset FLUSH_STORED_VERTICES in _tnl_execute_cassette as this may
      drawarrayrange fix
      Test programs for ycbcr extension
      move LoadYUVImage into readtex.c
      Merge Jose's documentation and core Mesa changes from embedded branch
      Fix typo (& instead of &&) to fix olympic.c bug
      Fix samples/fog.c regression - missing test for NeedEyeCoords.
      Store material attributes in an Attrib[] style array.  This is a
      Newer version of this file
      miniglx test programs
      r200 driver, brought over by Jon Smirl
      mga driver, brought over by Jon Smirl
      Shared radeon/r200 init code
      Not-quite-functional swrast-only dri driver
      Shared dri driver resources
      Use correct conversions when translating array colors.
      Fix typo
      VertexAttrib4fv wasn't being installed (4f was).
      index wasn't being calculated correctly (result was being discarded)
      Eric's mesa-depth32.diff
      build fixes, rh9
      Remove a lot of rcs tags, avoid merge conflicts
      Import vtx-0-2-branch
      Merge vtx-0-2-branch
      remove dead files
      Save and restore front colors on backfacing triangles rather than trying to
      Ensure PRIM_BEGIN marker isn't lost when incomplete primitives are wrapped.
      Backout last change on the displaylist path to fix gears (and probably
      Relax assertion to account for empty primitives.
      Remove PRIM_PARITY reference
      get tdfx compiling
      get ffb compiling
      Can't include <stddef.h> when building XFree86 server - define
      Provide dummy definitions of __driConfig externs
      Remove $Id tag
      No need to call radeonUpdateMaterial() from RunPipeline().
      No need to call UpdateMaterial() from RunPipeline().
      clean up some compiler warnings
      Call UpdateMaterials when they have changed
      Fix incorrect removal of PRIM_PARITY code (ie remove the parity code, not the
      Fix VERT_SET_RGBA, VERT_SET_SPEC macros to account for change to floating
      Fix use of uninited value 'flags', removed unused vars.
      Fix VERT_SET_RGBA, VERT_SET_SPEC macros to account for change to floating
      Fix VERT_SET_RGBA, VERT_SET_SPEC macros to account for change to floating
      MGA only has one vertex format.
      Fix typo.
      Remove old ubyte-color support code.  Update interp, copy_pv for float
      In fact, bring all the changes from mesa/tnl_dd to mesa/drivers/common.
      Remove dead file glcore-new.h
      Quash annoying compiler warning
      Updates to tnl_dd_dmatmp.h
      Updates to tnl_dd_dmatmp.h
      add driClipRectToFramebuffer helper function
      Don't evaluate GET_CURRENT_VB_MAX_VERTS() until after calling INIT()
      set MaxTextureImageUnits, MaxTextureCoordUnits
      new files
      Correct sense of ctx->_ForceEyeCoords (Felix Kuhling)
      Don't allow lighting in model coordinates when fog is active,
      Re-commit t_vertex.[ch] changes to fd.o server.
      Add t_vertex.c
      Beef up t_vertex.c:
      Remove debug
      Slight cleanup
      Apply attribute offset when calling extract().
      Fixes for _swsetup_Translate()
      SSE segfaults in conform - disable for now.
      Add comment regarding sse disable
      Testing
      Perform check for dstclip[3] == 0.
      Roland Scheidegger's r200_changeemitorder.diff.
      Don't call _tnl_draw_range_elements() unless start == 0.
      Fix up more confusions with count vs end in array functions
      Respect the 'in_use' field in the global LRU as a marker for
      Migrate i830 driver to t_vertex.[ch] for building hw vertices.
      some more count vs. end confusion
      remove duplicate t_dd_* files
      Another mechanism to create vertices with holes - a new EMIT_PAD style
      Use new EMIT_PAD functionality for spec and/or fog as available.
      Minor tweaks to help out at a driver level.
      Don't set extra bits in FLUSH_VERTICES, fix several state bugs.
      Add tests for ARB_fp trig instructions and also for 1D texturing
      Fix for recursion in generic_copy_pv_extras.
      Tweak: set z scale to one, else tnl thinks we have a 3d texcoord.
      Consolidate texObj->Pos/Neg/X/Y/Z and texObj->Image into a single
      Catch and shortcircuit no-primitive and no-vertex cases in _tnl_wrap_buffers()
      Use NEED_SECONDARY_COLOR() for checking specular state at all times
      Move FRAG_BIT_ definitions to mtypes.h so that NEED_SECONDARY_COLOR() macro
      Add glutInitWindowPosition call
      add glutInitWindowPosition calls
      Fix extract_3f_xyw().
      Use quads instead of lines to ensure this is testing stencil functionality
      Fix off-by-one in calculations for wrapped trifan, polygon primitives
      Add comments to make decyphering the faces easier
      Make it easier for drivers to create "subclasses" of the existing
      Remove debug
      glutInitWindowPosition
      Add _tnl_set_attr() to complement existing get_attr() call.
      Streamline the error path in VertexAttrib functions.  Makes things
      Sketch of codegen templates for t_vtx_api.c, not complete
      First round of codegen for t_vtx_api.c -- ie the Begin/Vertex/End code.
      new files
      Accomodate ARB_fp XPD opcode separately from NV_fp's X2D.
      Amusing utility to print ARB_fp programs as C code
      Make the output a little more likely to compile...
      Why are there two copies of this file?
      dynfn --> _tnl_dynfn
      Remove test for USE_X86_ASM & HAVE_NONSTANDARD_GLAPIENTRY
      Rename _x86_ symbols to avoid conflict with r200 driver
      Build a single string containing the generated 'C' code.
      Don't try to use .c=.o rule on .S files when building mesa.a
      First pass at support for building DRI drivers in mesa tree.
      First attempt at building a dri module in this tree
      Remove mesa.a when making clean
      Link against GL_LIB_DEPS
      Hack to get things compiling - include "drm.h" instead of "xf86drm.h".
      Fail if any subdir fails.
      Use tcc and the emitted C code from s_fragprog_to_c.c to dynamically compile
      new file
      Make _SAT instructions compile correctly.
      Make sure ctx->Driver.NeedFlush is set for Eval-generated attributes.
      When upgrading copied vertices, distinguish between an attribute
      Simplify last fix slightly, apply to vtx paths as well.
      Add printing for recently added opcodes.
      Fix order of EMIT_4UB_4F declarations
      Add fps output
      Allow *_dri.so to build in Mesa tree with the 'linux-dri' target.
      AMD64 changes from Ronny V. Vindenes.
      Define GLX_DIRECT_RENDERING in Makefile.template
      Add EXT_vertex_cull support to mesa
      Fix problems when sizeof(Node) != sizeof(float)
      New driver for i915 as well as older i830/i845/i865 chipsets.
      Fix batch.ptr when dropping a batchbuffer segment due to zero cliprects.
      add missing license texts
      Use correct macros for packing pixel colors -- this bug is probably
      Add infrastructure for t_vertex.c codegen.  Add an example driver
      Add t_context.h include for drivers.
      Rename 'emit_func' usages in drivers.  Will also rename the ones in
      Remove __driRegisterExtensions
      Rename the various function types in t_context.h to include a tnl_ prefix.
      The i915 driver was merged after EMIT_*_BGR bug was fixed.  This brings
      r200-maybe-flush-less-3.diff
      Add a license to this file.  It was written on VA's watch, so they get
      Nicolai's sw-clipspan-fixes.patch
      Update expected DRM version to 2.0.0
      EXT_422_pixels test
      Demonstrate using a color matrix for yuv-to-rgb conversion at texture
      Improve the behaviour of the build system wrt depend files.
      Get linux-solo dependencies building correctly,
      Fix a couple of glitches
      Add some catchall commands to realclean target to remove built files
      Add _mesa_delete_texture_object() call to viaDeleteTexture
      Remove #ifdef DEBUG's in code, but still allow compiler to remove debug
      Add a -fullscreen option
      Add vsync swapbuffers.  This waits on the irq so gears run in this mode
      Add missing swap-control calculations.
      Note that state is dirty on contended lock.  Allows two applications
      Push a number of global variables into the viaContext struct.
      Remove debug code which referenced an old global variable.
      Remove dead code.
      Fix merge error.
      Chop out more dead code.
      Remove the VIA_PERFORMANCE code.  A step towards moving the driver
      Large update
      Get twoside-lit triangles working again
      Fallback on 3d textures correctly.
      Don't pingpong cliprects through sarea on CopyBuffer().
      Make line stipple a fallback.
      Don't advertise wide lines or points.
      use clamped Line._Width in calculations
      Use Point._Size in calculation
      Simplfy clear() and swapbuffers() code.
      Fix some wrapping bugs in the last commit.  Probably there are more
      Fix MagFilter state for texunit 0
      fix regression with fallbacks from recent commits
      Don't need to adjust for drawXoff in fallback clipping.
      Initialize depth pointer correctly.
      Simplify viaBlit a bit more.
      Calculate DEPTH_SCALE correctly for polygon offset.
      Get scissor test working again.  Passes glean scissor test.
      Change to use the t_vertex.c mechanisms for building vertices,
      Turn specular lighting state on/off appropriately.
      Add GL_EXT_secondary_color
      Add GL_EXT_fog_coord
      If there is only one normal, don't transform it multiple times.
      Add a couple of hardwired fastpaths to t_vertex.c.
      Avoid a segfault in multiarb.c
      make alloc-dma functions inline, rearrange some debug
      Give attributes with zero-stride a count of 1 to make it easier
      Improve some quad paths, gives gears a little boost.
      Make the format of the fastpaths cleaner at the expense of a little
      Redraw on keypress, as apparently intended.
      Make clean in the x86 directory.
      Add an install target
      Improve the semantics of the 'install' target to make sure the right
      Add a noop install target
      Add a clean target to remove generated files.
      Remove -C flags for install which aren't universally supported.
      Use the 'to' vector to determine the loop count.
      Cope with the possibility that incoming vectors may have
      Add some more hardcoded fastpaths.
      Templatize the fastpaths.
      And fix the obvious bugs in higher-numbered templates.
      Add t_vb_cull.c (Bernard Blackham)
      Fix segfault in pipes by dealing with stride == 0 case in generic_interp_extras
      More changes to cope with color stride == 0
      Add a big nasty fallback for AlphaTest -- seems to always be wrong
      Temporarily disable pageflipping assert.
      Disable bogus pbuffer code (though this could be revived with a little
      Allow two GL primitives to be combined to one HW primitive
      Don't emit HC_SubA_HSPXYOS on CLE266.
      Add flag to clear texture caches after texture upload.
      Expand out the N(x) macro so that tags can find functions like
      new test to exercise context and window create/delete
      Remove bogus write to freed memory (valgrind).
      free buffer in t_vertex_c.c on context delete
      Simplify usage of drmHash functions and fix bug in
      Don't special-case 16bpp for colormask register - not needed.
      cut out dead texture state code
      Emit constant colors for texunit 1 blending.
      Typo
      Remove misleading comment.
      Simplify viaChooseVertexState slightly.
      Rename via_fastrender_* to avoid confusion with similar functions in
      Fix loop for quads-as-triangles.
      Populate WriteMonoDepthSpan in 24/8 modes.
      Fix previously un-noticed issue with flat-shaded points.
      Fix calculation of last_count in _tnl_wrap_buffers()
      Depend tweaks
      Remove depend on make clean
      Always use ARGB8888 hardware texture format when mesa specifies
      Bring the texcombine fixes to the trunk.
      Determine ahead of time whether a display list will include vertices
      GL_ARB_texture_rectangle enums (Andreas Stenglein)
      More texture enums (Andreas Stenglein)
      argb8888 optimizations from via branch
      SUBPIXEL fixes from Dieter
      Catch no-op vertex buffers consisting only of vertices which will
      mesa-tnl-0-to-NULL.patch from Jeff Muizelaar
      mesa-main-0-NULL.patch from Jeff Muizelaar
      Turn _via_fastrender_stage back on.
      mesa-fix-one-bit-signed-bitfield.patch from Jeff Muizelaar

Kendall Bennett (16):
      . Updated GL/gl.h with GLCALLACKP and GLAPIENTRYP macros for compatibility
      . Updates to get X86, MXX and 3DNow assembler code working with Watcom
      . More fixed to
      . More Watcom C++ warnings fixed in Mesa.
      Updates to fix problems building with the Open Watcom compiler.
      Updates required for building the SciTech SNAP version of Mesa. This includes:
      Fixed bugs in Mesa software span rendering for color index modes
      Fixed incorrect Watcom assembler macros
      Removed bogus OS/2 floating point->integer conversion routine
      More SciTech SNAP specific changes to glheader.h. Also modified the macros
      Added missing #undef SPAN_VARS to swrast/s_spantemp.h
      Updates to x86 assembler support for Open Watcom and SNAP
      Updates to SGI GLU code to get it to compile clean with the Open Watcom compiler.
      Added GLAPIENTRY decorations for all first level OpenGL API function entry
      Updates to SSE assembler support for Open Watcom
      More SciTech SNAP updates. Some of these files didn't really change, but

Kevin E. Martin (1):
      fixed SVR4 'cc' compiler macro problem (dawes at xfree86.org)

Michal Krol (71):
      Initial registry for ARB_fragment_program and ARB_vertex_program.
      Added support for ARB_fragment_program_shadow.
      Grammar package supporting 8-bit registers.
      Added missing Copyrights.
      Initial changes after moving from src/mesa/main/.
      Fixed a number of typos.
      Moved from src/mesa/main.
      Slang internal include file defining constructors and operators
      The OpenGL Shading Language Runtime Library for Legacy Target.
      Moved to src/mesa/shader
      Moved to src/mesa/shader and renamed to arbprogparse.[ch]
      Moved to src/mesa/shader and renamed to arbprogram_syn.h
      This wont be needed any more
      Moved from src/mesa/main
      Enable draw_buffers only if GL_ARB_draw_buffers string is
      Prefix operator and constructor keywords with two consequtive underscores
      conform to language version 1.10 rev 59
      conform to shader spec 1.10.59
      - speed up syntax parsing - the parser is now 4 times faster
      add new grammar_check_fast entry for faster parsing
      change program options numbering to accommodate future
      remove redundant check of parsed program target
      - use new program option values from arbprogram.syn
      forgot about revision increment
      - add decimal literal integer support
      by karoshi - add support for ARB_shader objects and ARB_vertex_shader
      the file was empty!
      fix DrawBuffersARB entry
      fix DrawBufferARB entry
      regenerate with correct script
      regenerate with correct script
      assign api offsets for ARB_shader_objects and ARB_vertex_shader
      add support for ARB_shader_objects and ARB_vertex_shader
      add support for ARB_shader_objects and ARB_vertex_shader
      by karoshi - header file for ARB_shader_objects implementation
      by karoshi - ARB_shader_objects implementation
      add FEATURE flags for ARB_shader_objects, ARB_vertex_shader and
      add entries for extensions ARB_shader_objects, ARB_fragment_shader and
      fill api entries for ARB_shader_objects and ARB_vertex_shader
      remove unused entries
      renumber ARB_shader_objects and ARB_vertex_shader offsets
      slang syntax conforming spec rev. 1.10.59
      aliasing was broken
      add shader/shaderobjects_3dlabs.c
      init ARB_shader_objects subsystem
      enable ARB_fragment/vertex_shader
      add ARB_shader_objects interfaces, shared and context state
      add shaderobjects_3dlabs.c
      add atifragshader.c, shaderobjects.c, shaderobjects_3dlabs.c, atifragshader.h,
      implement ARB_shader_objects, not complete
      cosmetic changes
      3dlabs glslang frontend, not complete
      3dlabs glslang frontend
      add s_atifragshader.c, s_atifragshader.h
      add src/mesa/shader/slang to INCLUDE_DIRS
      add [.slang] to INCDIR, fix comma in OBJECTS
      add 3dlabs_shhandle interface
      slang to mesa compatibility, tested only under Windows
      2005-oct-19 3dlabs slang compiler, tweaked a little bit
      add slang/ directory tree
      fix MAX_VERTEX_ATTRIBS_ARB query
      fix // comments
      add slang preprocessor C sources; add slang include directories
      change .cc suffix to .cpp
      add shader/slang/slang_mesa.cpp
      comment out _THROW calls, dont want to compile under Linux
      remove #ifndef linux, it isnt defined on Linux, where should I define it?
      Handy TIntermTraverser class wrapper
      add slang/traverse_wrap.h
      add getstring.c; remove arb*.h nv*.h stuff
      move file

Michel Dänzer (12):
      Actually disable HW TCL capability when SW TCL is forced.
      Fix initialisation of cubic tecture offsets (Andreas Stenglein)
      Fix cosmetic GL_RENDERER string problem (Andreas Stenglein)
      Fix attenuation hardware state handling, inspired by the R200 DDK reference
      Call UpdateMaterial() unconditionally on GL_COLOR_MATERIAL state change
      Fix hardware ROP state handling (Roland Scheidegger)
      Only print message about disabling HW TCL if the chip supports it in the
      Inline get_size as an interim measure to get rid of function call overhead.
      Call _mesa_ResizeBuffersMESA() in r200Viewport(), fixes incorrect rendering
      MMIO endianness fixes for Radeon drivers.
      Typo fix.
      fd.o bug #2576: Add support for ATI RN50/ES1000. (ATI Technologies Inc.)

Miklos Fazekas (11):
      More compliant error check.
      Minor cleanup.
      Added support for GL_ext_texture_add
      Packed RGB support.
      Packed RGBA support.
      Added packed ARGB support.
      Added gl_ext_multitexture_ADD suport.
      Added PARB support.
      Added glGetMinmax
      Fixed a MacOS specific compile error.
      Added glColorTable, etc.

Nicolai Haehnle (17):
      Initial revision
      Huge dumb drop. State:
      - FIX: flickering
      Hardware accelerated depth clear
      Implement ColorMask
      Add culling registers, sync versions of r300_reg.h
      No visible changes, but commit the groundwork for further experiments:
      Synchronize r300_reg.h.
      Fix more warnings and compilation issues.
      Remove unused variables from r300_context
      RADEON_DEBUG is a bitfield, so use & instead of ==.
      Fixing *lots* of warning messages, especially concerning unportable
      Fix: RADEON_DEBUG is a bitfield, so use & to check for flags, not ==
      Reset nr_released_bufs to 0 after the command buffer has been submitted.
      Always submit command buffers, even when there are no cliprects,
      r300RefillCurrentDmaRegion: Retry buffer acquisition only once, after
      Print an error if idling the engine before the buffer copy fails.

Philippe Houdoin (34):
      Add a BeOS build config.
      Move BeOS build into main Makefiles chain.
      Fix a bug that set RGBA bits length always to 0 (reported by Marcin Konicki).
      BeOS's _glthread_DECLARE_STATIC_MUTEX() couldn't be used outside a function
      Silent compiler "variable may be used uninitialized" warnings.
      Obsoleted. Now, the BeOS targets build by main Makefiles chain.
      Fix demo.cpp, which wasn't working as expected.
      Obsolete BeOS-specific makefile.
      Obsolete the BeOS-specific makefile as we now build using standard Makefiles chain.
      Updated according to new way to build under BeOS.
      BeOS now officially moved to main Makefiles chain. To build under BeOS: "make beos"
      Update BeOS mklib to set mimetype and version.
      Move to $(TOP)/configs/default-based makefile.
      Update to Jake Hamby's BeOS GLUT 3.7 port.
      Replaced the .cpp-ized (for build issue) common GLUT source files by their
      Replaced the .cpp-ized (for build issue) common GLUT source files by their
      Remove obsoleted BeOS-specific makefile as the main Makefile system now
      Forgot to commit this valuable file! :-)
      Add a guard to stop building libGLU.so under BeOS, as it's build into libGL.so.
      Remove obsoleted BeOS-specific makefiles.
      Turn gcc optimization up to -O3.
      Obsoleted as demos programs build on BeOS with standard Makefile system.
      GLInfo now show GLU and GLUT infos too. Fix the includes paths
      Clearer guard message text.
      Stupid me. I don't know why I moved the GLView.h here at first, but now I know why
      make clean at root now clean the glu/$(GLU_DIRS) too.
      Add glutGetProcAddress(). Not all GLUT version 5 APIs are supported yet, thought.
      Duplicate CFLAGS in CXXFLAGS removed. Now define DEBUG symbol in debug build.
      Todo list updated. Build instructions detailed.
      Disable x86 iround() version on BeOS with gcc < 2.95.x, as it choke GAS.
      Implement BGLView::CopyPixelsIn/Out().
      Forgot a contributor.
      Fix a potential memory leak.
      Let's build again after the support for GL_ARB_draw_buffers was added.

Randy Frank (5):
      Added test for the occlusion test code.
      Added entry point and define for the occlusion test.
      Added occlusion test code.  Follows the HP proposed occlusion test extension.
      Updated to compile under Irix 6.5
      Implemented support for the HP occlusion test extension (osmesa and X)

Roland Scheidegger (53):
      remove bogus check of texture target in RadeonTexParameter
      fix potential problem overwriting dma region
      align blits for texture rectangles correctly, fix potential problem overwriting dma region
      remove bogus check of texture target in R200TexParameter
      inform driver of changed wrap/filter parameters due to binding of nv_texrect texture
      Changed lighting to use SOURCE_MATERIAL instead of PREMULT (moves some light color calculations to the GPU). Two-side lighting fallback no longer needed and eliminated.
      change lighting to use MULT instead of PREMULT (i.e. move some lighting color calculations to the GPU)
      revert to version 1.18 for now to fix segfaults in some applications
      texture env combine refactor (Andreas Stenglein), following the same principles as the refactored code already present in the R200 driver
      Add support for the new DRI driver interfaces to the radeon driver (patch ported from the r200 driver, r200_screen.c rev. 1.8 - 1.12)
      Fix calling fallback_drawarrays() with the wrong arguments
      fix bracket at wrong place
      Fix GL_MIN and GL_MAX blend equations (set blend factors accordingly). Fix errors when blending is disabled (set blend equation and function to default values).
      add missing R200_RB3D_BLENDCOLOR register
      add check if drm is new enough for packet R200_EMIT_RB3D_BLENDCOLOR
      Add support for GL_EXT_blend_[func|equation]_separate. Fix GL_EXT_blend_color. Remove support for GL_EXT_blend_logic_op (cannot be supported together with blend_equation_separate unless a software fallback would be added).
      fix constant texture environment color setup for texture units > 0
      (Dieter Nuetzel) fix typo in help screen
      use OGL 1.1 color logic ops in addition to GL_EXT_blend_logic_op
      change some explicit references to texture units 0/1 to unit < MaxTextureUnits (Andreas Stenglein, Ronny Vindenes)
      remove never finished verts path (r200_maos_vbtmp.h, r200_maos_verts.c) and unused R200_OLD_PACKET macro
      update so it compiles again (untested driver path!). Basically ported changes from r200_maos_arrays.c cvs rev. 1.3-1.5.
      add support for GL_ARB_texture_env_crossbar
      change GL_ARB_texture_env_crossbar implementation behaviour to OGL 1.4 core specification instead of extension specification (undefined behaviour instead of disabled blending if not enabled texture units are referenced). Fix minor signedness bug.
      add support for more than 2 texture units (max 6, default 4). use hang workarounds only for r200, not derivatives.
      fix minor signedness issue in assertion
      added more compressed formats to is_color_format()
      fix tex env mode GL_BLEND for GL_INTENSITY texture formats
      use I8 internal format for GL_INTENSITY, GL_ALPHA and GL_LUMINANCE texture formats. Fix potential bug with tex env combine mode GL_DOT3_RGB/GL_DOT3_RGB_EXT and ALPHA_SCALE
      use I8 internal format for GL_INTENSITY, GL_ALPHA and GL_LUMINANCE texture formats. Fix various small bugs with tex env combine mode GL_DOT3_RGB(A)/GL_DOT3_RGB(A)_EXT
      revert accidental change from rev. 1.10
      new test which uses mixed texgen/non-texgen texture coordinates to exhibit potential bugs in hardware drivers
      add missing R200_RB3D_BLENDCOLOR to r200/radeon_sanity.c
      correct X86_DISPATCH_FUNCTION_SIZE if THREADS is defined
      fix incorrect overflow detection when emitting R200_SS_MAT_1_SHININESS
      enable GL_EXT_fog_coord. Calculate fog factors and submit them instead of fog coords (it seems the chip cannot do fog factor computation when not using fragment depth as fog coord source). vtxfmt uses fallback for now (most code present but some magic would be needed if replaying vertices is necessary later on).
      add some chip ids
      fix quads to use GL_QUADS instead of GL_TRIANGLES as primitive
      (from r200 driver) If an application cleared before any state had been emitted, that clear would
      add LOCK_HARDWARE/UNLOCK_HARDWARE to radeonClear so state gets updated which hopefully fixes lockups (?) (from r200 driver)
      enable hw quad primitive for swtcl
      revert enabling of hw quads for swtcl. It lead to problems with quad_strips, since they'll get reduced to quads sometimes if hw quads are enabled. But this needs more thought, since it looks like clipped primitives will always be emitted as polys, so the reduced primitives for triangles, quads, quads_strips, polys all need to be the same, otherwise get lockups with for instance the olympic test. Render templates would probably need to be changed for this to work.
      enable GL_EXT_stencil_wrap (patch from idr), including some hacks for original radeons which have some broken stencil ops.
      only enable GL_MESA_ycbcr_texture for real r200 chips, not the derivatives, since yuv textures do not work for some reason on the other chips.
      fix depth/stencil readback if coordinates are larger than 1023 (https://bugs.freedesktop.org/show_bug.cgi?id=2010). Should now work up to 2047, which is the current limit for 3d rendering.
      (Stephane Marchesin, me) add hyperz support to radeon and r200 drivers. Only fast z clear and z buffer compression are supported for now, hierarchical-z is not. Still problems with multiple apps and z/stencil readback, which is why hyperz is disabled per default. Also add the new point sprite packet drm 1.13 accepts to the sanity code.
      replace magic number with macro constant RADEON_ZBLOCK16
      new version check to allow to test for a range of ddx major versions, instead of just one major version
      (Stephane Marchesin, me) Add support for color (framebuffer) tiling to the radeon and r200 driver
      (Andreas Stenglein) fix projected textures with swtcl, they need the w coordinate of the vertex (bugzilla #1648)
      add the new RADEON_CUBIC_OFFSET/FACES state packets/registers to radeon (and r200) sanity code
      add texture micro and macro tiling to radeon/r200 driver. This can improve performance up to 15% in texture-intensive applications. Convert the driver to use the correct blit format and blit width instead of fixed blit format and blit width when uploading textures to make it work.
      remove accidentally commited printf for tiling support

Rune Petersen (2):
      initial r400 support: r400 cards are treated as RV350 for now.
      small cleanup, remove some warnings

Ted Jump (20):
      Update for beta3
      Added ESC key handling
      Cleaned up exit from main to prevent MSVC warnings
      I changed the xxxGetProcAddressEXT function declarations to be more
      Win32 build req't updates
      Patch error fixup
      wgl function prototypes moved out to better work with fxwgl
      Relocated wgl function prototypes to separate .h for easier work with fxwgl & glut
      Modified to build better with Mesa wgl functions
      More Win32 build compliance fixups
      Added NASM support
      Removed use of GLU.H - prevents compiler warning wrt:gluGetProcAddressEXT
      Fixed LIBNAME entry
      Changed name 'glGetProcAddressEXT' to 'gluGetProcAddressEXT'
      Fixed parameter type for notify function pointer parameter to gl_extensions_add
      Minor patches to cleanup some compiler warnings in MSVC
      Added pragma to disable warning about unk pragmas for MSVC5
      Added checking for FXMESA_USE_ARGB to disable pixel reformatting when not appropriate
      Changed over to maximal warnings on Win32 builds.
      turned off 'conditional expression is constant' warning for win32

Thomas Hellström (6):
      Unichrome DRI driver:
      Unichrome DRI:
      Unichrome DRI:
      Unichrome DRI:
      Unichrome DRI driver:
      Small Unichrome fixes:

Ville Syrjala (2):
      Fixed off by one errors in clipping.
      Removed two unnecessary variables.

Vladimir Dergachev (86):
      The types uint32_t and uint8_t are not used in Mesa tree.
      Fix cut and paste error (reported by Jerome Glisse)
      Update with most recent version.
      Expose primitive types being rendered, in preparation to implement fixed pipeline primitive drawing.
      Implement rendering of (flat color) QUAD primitives as an experiment.
      Use R300_CMD_END3D for end_3d().
      Bring in latest revision of r300_lib.
      Get most primitives working using immediate mode.
      Rearrange code so we don't dump state as often.
      Port viewport setting code from R200.
      Clarify some of the unkXXXX atoms.
      Change default RS settings so that glxgears will display something..
      Restructure code.
      Hooked up projection matrix - the gears actually rotate !
      Update to recent r300_lib.
      Update vertex buffer code (still does not work properly..)
      Add missing files.
      Rework slightly r300_get_primitive_type - make it clearer and more compact..
      Add texture drawing code. Note: it is broken at the moment and is disabled in CVS. However, all hooks are there.
      Sync with master copy.
      Qualify the magic 20B0 register as SE_VTE_CNTL similar to R200. Looks like disabling Z offset and scaling displays gears properly, not sure why. Perhaps the meaning of the bits has changed ?
      Add a convenience function to issue CP delays.
      Get textures to work with NeHe lesson06.
      I think I fixed the lockups issue.
      Cleanup !
      Add texture units registers
      Port texture allocation code from R200.
      Enable code to compute other constants that were used by R200 code.
      Turns out I left flat primitives in vertex buffer mode. Switch them back to immediate which works correctly..
      Add structs describing vertex and pixel shader state to r300_state.
      Major code restructuring:
      Unify rendering of textured and non-textured primitives.
      Port code from r200 that implements color blending. Seems to work.
      Sync with master copy.
      Play a little bit with texture formats.
      Reduce the amount of debug output, while still printing important messages.
      Switch to a more complicated scheme of choosing texture formats, as it looks
      Add texture format to get lesson19 working again.
      Turns out the kfountain.kss (as well as lesson19) had a bug - the size was set wrong. Fixed.
      Implement a more elaborate hashing scheme for texture formats. Still not perfect.
      Port EmitPixelShader from cmdbuf.c to SetupPixelShader in state.c.
      Clean up the code and make it unnecessary to issue extra calls to Setup*Shaders
      For some reason we need r300Flush when using textures. Perhaps the problem is
      Bring vertex buffer code up to date..
      Copy sample pixel and vertex shaders to r300_fixed_pipelines.h
      Switch r300_state.c to use r300_fixed_pipelines.h instead of r300_lib.h.
      Move the few definitions from r300_lib.h still in use to r300_emit.h.
      Switch back to using immediate mode code - use of AGP space is interfering with texture management.
      Get NeHe lesson08 to work. Note: it appears the filtering does not work properly. Need to find place where it is broken and put an fprintf() complaining about it.
      Incorporate TX_FORMAT information contributed by Wladimir van der Laan.
      Prevent "disabling 3D acceleration" message from printing - we do not define raster functions anyway.
      Sync with master copy.
      Incorporate latest data from Wladimir van der Laan.
      Reduce amount of stderr output.
      Remove redundant register.
      Use symbolic constants to describe pixel shader ALU instructions.
      On the way to getting stencil and texture formats working.
      On the way to getting stencil working.
      Properly input stencil info contributed by Wladimir.
      Properly configure routing - based on advice by Keith Whitwell.
      Remove lots of old code that dealed with "magic" texture formats.
      struct ati_fragment_shader was not defined anywhere, define it to get r300_vertexprog.c to compile.
      My bad, the old code works fine if updated to newer Mesa tree.
      Use SecondaryColorPtr, not ColorPtr[1] (the latter is NULL).
      Add magic sequence to prevent ClearBuffer from locking up.
      It helps to print actual values instead of random ones.
      Looks like we can define FIRE_VERTICES correctly.
      Properly set ABLEND and CBLEND registers. The older code worked, but by accident.
      Get alpha in textures to work properly.
      Reduce noiseness of the driver.
      Make WARN_ONCE messages more informative.
      Get Polygon offset fill to work.
      This was too optimistic sorry - all that register does is something weird with polygon culling.
      Mark ZBIAS definitions.
      Remove redundant instructions from fixed pipelines, looks like they do not do anything, at least on my hardware.
      Since high bits can be used for flags, mask them out.
      Get vertex programs supplied by the user work properly with immediate mode - vertex buffer mode worked by accident.
      Looks like we need at least one instruction for pixel shader.
      Forgot to increase the length.
      Start beginning of pixel shader generator..
      Warn of unknown texture formats.
      Let Polygon Offset be turned on and off.
      Remove #ifdef's around PolygonOffset function.
      Fix copy'n'paste errors.
      Cleanup formatting a little bit.
      Add code to autogenerate vertex shaders.

davem69 (7):
      Add USE_SPARC_ASM diagnostics to print_info.
      Python script for making glapi_sparc.S
      Generate this now that I know which
      Sparc optimized GLAPI dispatch table.
      SPARC normal tnl optimizations.
      Remove dependency on MESA sources from gl* files
      Use #function instead of @function

hmarson (2):
      Added 0x4e54 (Thinkpad r50p) as PCI_CHIP_RV350_NP
      Added 0x4e51 - Saphire 9600 256MB

jtg (1):
      Initial revision

pesco (8):
      * include/GL/Makefile.am: ENABLE_GLUT_BUILD -> NEED_GLUT
      * demos/Makefile.am (INCLUDES): Added -I$(top_srcdir)/util.
      Forgot to add newly created files.
      * gen_srclists.sh: Use getopt to parse the command line. Added command
      * si-glu/libnurbs/: Undid move of glimports.h, mystdio.h, mystdlib.h.
      * common_rules.make: Created to hold the conditional build rule
      * src/swrast_setup/ss_tritmp.h (quad): Changed name of this function to
      More .cvsignore updates.

reist (1):
      get two textures to work

sio (1):
      Added missing GLAPIENTRY to glConvolution* and glCopy*

sio2 (1):
      Added GL_TEXTURE_PRIORITY to gl_TexParameterfv in src\texstate.c

tanner (4):
      merged the autoconf build system from experimental-1 branch
      added installation instructions
      lots of autoconf updates
      include conf.h



More information about the Xquartz-changes mailing list