[cairo] glitz, pbuffers, and OpenGL 2.0

[David Simmons]

Your modification fixes one part of the problem -- the GLX 1.2/1.3 
issue.  However, I still can't use pbuffers because the pbuffer 
extension is not being advertised and flagged as a feature in glitz.  If 
I add a few lines to force the fbconfig and pbuffer flags to be set, as 
shown below, pbuffers work fine.

David

diff -Nru glitz-20060213.orig/src/glx/glitz_glx_extension.c 
glitz-20060213/src/glx/glitz_glx_extension.c
--- glitz-20060213.orig/src/glx/glitz_glx_extension.c   2006-02-10 
02:04:32.000000000 -0700
+++ glitz-20060213/src/glx/glitz_glx_extension.c        2006-02-13 
19:41:37.000000000 -0700
@@ -73,6 +73,11 @@
            /* ATI's driver emulates GLX 1.3 support */
            if (!strncmp ("ATI", vendor, 3))
                screen_info->glx_version = 1.3f;
+
+           /* ATI's latest driver supports fbconfig and pbuffer,
+              although they are not advertised as extensions. */
+           screen_info->glx_feature_mask |= 
GLITZ_GLX_FEATURE_FBCONFIG_MASK;
+           screen_info->glx_feature_mask |= GLITZ_GLX_FEATURE_PBUFFER_MASK;
        }
     }
 }


David Reveman wrote:
> Hey,
>
> I know that ati's driver has support for pbuffers even though they don't
> have server side GLX 1.3. I've committed some code that checks if vendor
> string starts with ATI and assumes GLX 1.3 if that's true. I haven't
> tested it yet, let me know if it doesn't work for you.
>
> -David
>
> On Wed, 2006-02-08 at 08:25 -0700, David Simmons wrote:
>   
>> [ From my study of the archives, I'm assuming that it's okay to discuss 
>> glitz on this list. ]
>>
>> Hi,
>>
>> I'm using a version of glitz that I checked out from CVS on Jan 16, 
>> 2006, and it doesn't seem to provide pbuffer support on my OpenGL 2.0 
>> platform -- an ATI Radeon 9550 running in Linux using the ATI fglrx 
>> 8.21.7 driver.  ATI's sample fgl_glxgears pbuffer demo runs without 
>> complaint.
>>
>> After some research and testing, my hypothesis is that OpenGL 2.0 / GLX 
>> 1.3 does not advertise the GLX_SGIX_pbuffer and GLX_SGIX_fbconfig 
>> extensions because they are now part of the standard.  When glitz does 
>> not detect these extensions, it disables these features.  (I haven't 
>> studied the OpenGL/GLX specifications yet to see if this is indeed the 
>> proper behavior.)
>>
>> I noticed that the SDL folks seem to agree with this hypothesis, as they 
>> made the following change to their code at some point:
>> -    if ( ExtensionSupported("GLX_SGIX_fbconfig", glx_extensions) &&
>> -         ExtensionSupported("GLX_SGIX_pbuffer", glx_extensions) &&
>> +    if ( ((this->gl_data->glx_version >= MAKE_GLX_VERSION(1, 3)) ||
>> +          (ExtensionSupported("GLX_SGIX_fbconfig", glx_extensions) &&
>> +           ExtensionSupported("GLX_SGIX_pbuffer", glx_extensions))) &&
>>
>> I made two quick'n'nasty hacks to my glitz source to test this, and now 
>> pbuffers seem to be working properly and hardware accelerated:
>>
>> 1. glitz_glx_extension.c glitz_glx_query_extensions() -- I hard-coded 
>> the GLITZ_GLX_FEATURE_FBCONFIG_MASK and GLITZ_GLX_FEATURE_PBUFFER_MASK 
>> features into glx_feature_mask.
>> 2. glitz_glx_info.c _glitz_glx_proc_address_lookup() -- I hard-coded my 
>> glx_version to be 1.3.  (for whatever reason, glxinfo reports my client 
>> glx version as 1.3, but my server glx version as 1.2.  No matter... 
>> fbconfig/pbuffer is present, really!)
>>
>> David
>>
>> _______________________________________________
>> cairo mailing list
>> cairo at cairographics.org
>> http://cairographics.org/cgi-bin/mailman/listinfo/cairo
>>     
>
>
>