Hi, IMHO, It is valid to report error after glEndList with offend API removed from the list, and it is also reasonable to still keep the APIs and report the same error again when glCallLists. I have filed an BUG#12012 in khronos and regarding to the feedback I have collected, it seems we need to relax this test case.
If no objection, Crystal could you prepare a patch to address this issue? Thanks. Best Regards, David From: Piglit [mailto:[email protected]] On Behalf Of Jin, Crystal Sent: Friday, March 21, 2014 6:29 PM To: [email protected] Subject: [Piglit] whether require glBegin/glEnd check during display list compiling Hi, When I tested piglit test case 'spec/!OpenGL 1.0/gl-1.0-beginend-coverage', I found that mostly errors with our amd driver is expected error not match getError after 'glCallList(dlist)'. In another words, display list executed got error. And I found the root cause is that there is glBegin/End check functionality in our driver which leads to display list compiling directly return rather than add the GL command into display list. So the display list only includes glBegin and glEnd function. And my question is whether require glBegin/glEnd check during display list compiling, and need add wrong GL commands sequence into display list or not (our driver does not add it). For example, glBitmap cannot be in glBegin/glEnd, but it is in glBegin/glEnd during glNewList/glEndList, if it is required to add it into display list. Spec 4 page 411 "After calling NewList all subsequent GL commands are placed in the display list (in the order the commands are issued) until a call to Endlist(void)". It is not obviously mentioned whether it need to be added into display list if subsequent GL commands are wrong. Can you please help me to confirm the problem, thanks? Best wishes! Crystal Jin
_______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
