[GEM-dev] [ pd-gem-Feature Requests-2612731 ] glsl_geometry

[IOhannes m zmoelnig]

cyrille henry wrote:
> 
> 
> Matthias Neuenhofer a écrit :
>>
> 
>>> did you recompile Gem after my yesterday commit?
>> yes but with change
>> void glsl_program:: outtypeMess(GLuint outtype) {
>>   m_geoOutType=outtype;
>>   if(m_program && glProgramParameteriEXT) {
>>     
>> glProgramParameteriEXT(m_program,GL_GEOMETRY_OUTPUT_TYPE_EXT,m_geoOutType); 
>>
>>   }
>> was before GL_GEOMETRY_INPUT_TYPE_EXT
> 
> yep, your right, there is another typo here.
> i corrected it.
> it is strange it was working with this typo.
> now that i corrected it, nothing changed...
> 


that's because (i think) you cannot really change the in/output geometry 
once a program is linked (at least it doesn't seem to do much).
when the program get's re/linked, glProgParamiEXT gets called with the 
correct GL_GEOMETRY_... so that's why it should work anyhow (once the 
m_geoOutType/m_geoInType typos are fixed ...)

anyhow, i checked in a more-typo-fixed version.

> 
> so sending [draw foo( to the square will draw the square with GL_POLYGON...
> sending draw QUAD_STRIP will draw the square with GL_QUADS.
> sending draw Line / line_loop / LINE_STRIP will use gl_line_loop
> draw TRIANGLE_STRIP will use gl_triangle, you have to send draw strip in 
> order to have triangle strip.
> etc.
> so the drawing mode and there comments in my initial patch was right.
> changing them result in unexpected behaviour.


we should check a (generic) way to automatically find out which drawing 
mode should be used.
it will definitely become a usability problem, if people have to read 
the Gem-sources to know which geometry type they are allowed to use in 
the geometry shaders. also the use of GL_LINE_STRIP vs "fill" is rather 
... arbitrary.


mfgasdr
IOhannes

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3636 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.puredata.info/pipermail/gem-dev/attachments/20090427/ffef5e7d/attachment.bin>