Re: Patch for bug #901422 (CLAMP_TO_EDGE)

[Frank C. <dev...@ch...>]

OK, that looks about right (for the record I get 1.3 on Mac OS X 
10.3.5). I had thought OpenGL on Mac OS X was up to 1.4 or better but 
apparently it differs depending on the card driver.

Regardless, the function below should work in all cases now - perhaps 
someone can check this in when they have a minute.

Frank.

On 30-Sep-04, at 3:39 AM, Daniele Cavallini wrote:

> I try function with Window 98 and Windows Xp
> In Window 98 I obtain:
> openGLVersion = "1.15"
> majorVers=1
> minorVers=15
>
> In Window Xp
> openGLVersion = "1.3.3966 winXP Release"
> majorVers=1
> minorVers=3
>
> Daniele
>
>
> At 20.26 29/09/04, you wrote:
>> Hmmm, GLUtils_CheckExtensions looks like it has some problems...
>>
>> First, the version check is flawed:
>>
>>         if ( (majorVers >= 1) && (minorVers >= 2) )
>>
>> would make more sense as:
>>
>>         if ( (majorVers > 1) || ((majorVers == 1) && (minorVers >= 
>> 2)) )
>>
>> Or else a version like "2.0" will get bypassed. The minor version can
>> easily be two digits though, and "15" shouldn't be greater than "2" in
>> this case, so it's broken anyway...
>>
>> Apple's glCheck code converts the GL_VERSION string into a BCD for
>> comparisons
>> <http://developer.apple.com/samplecode/Carbon_AGL_Full_Screen/ 
>> listing2.html> Quesa should probably do the same, something like 
>> this:
>>
>> ------
>> void
>> GLUtils_CheckExtensions( TQ3GLExtensions* featureFlags )
>> {
>>         const char*     openGLVersion = (const char*)glGetString( 
>> GL_VERSION );
>>         const char*     openGLExtensions = (const char*)glGetString( 
>> GL_EXTENSIONS
>> );
>>
>>         short j = 0;
>>         short shiftVal = 8;
>>         unsigned short glVersion = 0;
>>
>>         // Initialize to default value, all off.
>>         memset( featureFlags, 0, sizeof(TQ3GLExtensions) );
>>
>>         if (openGLVersion != NULL)
>>         {
>>                 // Get BCD version.
>>                 while (((openGLVersion[j] <= '9') && 
>> (openGLVersion[j] >= '0')) ||
>> (openGLVersion[j] == '.'))
>>                 {
>>                         if ((openGLVersion[j] <= '9') && 
>> (openGLVersion[j] >= '0'))
>>                         {
>>                                 glVersion += (openGLVersion[j] - '0') 
>> << shiftVal;
>>                                 shiftVal -= 4;
>>                         }
>>                         j++;
>>                 }
>>
>>                 // Check for extensions.
>>                 if ( glVersion >= 0x0120 ||
>>                         isOpenGLExtensionPresent( openGLExtensions,
>> "GL_EXT_separate_specular_color" ) )
>>                 {
>>                         featureFlags->separateSpecularColor = kQ3True;
>>                 }
>>
>>                 if ( glVersion >= 0x0120 ||
>>                         isOpenGLExtensionPresent( openGLExtensions,
>> "GL_EXT_texture_edge_clamp" ) ||
>>                         isOpenGLExtensionPresent( openGLExtensions,
>> "GL_SGIS_texture_edge_clamp" ) )
>>                 {
>>                         featureFlags->clampToEdge = kQ3True;
>>                 }
>>         }
>> }
>> ------
>>
>> Note: I added the GL_SGIS_texture_edge_clamp check as well. I've
>> updated my patch archive with these changes, but for some reason I'm
>> not convinced GL_VERSION is always correct (though Apple seems to
>> depend on it). What OS are you running that returns OpenGL 1.15?
>>
>> Thanks,
>> Frank.
>>
>> On 29-Sep-04, at 12:42 PM, Daniele Cavallini wrote:
>>
>>> I am using  OpenGL 1.15. The value is obtained from glGetString(
>>> GL_VERSION ).
>>> When exit from  GLUtils_CheckExtensions featureFlags->clampToEdge is
>>> set to kQ3True;
>>> I obtain:
>>> majorVers =  1
>>> minorVers = 15
>>> after the step
>>> int             numScanned = sscanf( openGLVersion, "%d.%d",
>>> &majorVers, &minorVers );
>>> It is Correct?
>>> Daniele
>>>
>>> At 18.06 29/09/04, you wrote:
>>>> No, that'll break it...
>>>>
>>>> The value will be false to start (the struct is zeroed in the
>>>> function) and is set to true if using OpenGL 1.2+. The extension may
>>>> not exist in an OpenGL 1.2+ implementation though, so you'll be
>>>> setting it back to false even though it can be handled just fine at
>>>> that point. The extension check is strictly for older OpenGL 
>>>> versions
>>>> before the ARB ratified the function.
>>>>
>>>> Frank.
>>>>
>>>> On 29-Sep-04, at 10:59 AM, Daniele Cavallini wrote:
>>>>
>>>>> Question:
>>>>> Is It better add this "else" in function "GLUtils_CheckExtensions"?
>>>>>
>>>>>  if (isOpenGLExtensionPresent( openGLExtensions,
>>>>> GL_EXT_texture_edge_clamp" ))
>>>>>         {
>>>>>                 featureFlags->clampToEdge = kQ3True;
>>>>>         }
>>>>>         else
>>>>>                 featureFlags->clampToEdge = kQ3False;
>>>>> Daniele
>>>>>
>>>>> At 01.50 29/09/04, you wrote:
>>>>>> Here: <http://chaoticbox.com/quesa901422.zip> is a small patch 
>>>>>> that
>>>>>> checks for CLAMP_TO_EDGE and uses it when available for texture
>>>>>> shaders
>>>>>> that have a kQ3ShaderUVBoundaryClamp setting. The original report 
>>>>>> is
>>>>>> here:
>>>>>> <http://sourceforge.net/tracker/index.php?  >>>> 
>>>>>> func=detail&aid=901422&group_id=45158&atid=442052>
>>>>>>
>>>>>> On a similar note; Is there a better way to submit patches? Do you
>>>>>> guys
>>>>>> prefer some sort of diff tool output? If so, please let me know...