Merge lp:~jammy-zhou/glcompbench/glcompbench into lp:glcompbench
Proposed by
Jammy Zhou
Status: | Rejected |
---|---|
Rejected by: | Jesse Barker |
Proposed branch: | lp:~jammy-zhou/glcompbench/glcompbench |
Merge into: | lp:glcompbench |
Diff against target: |
155 lines (+42/-41) 3 files modified
src/libmatrix/mat.cc (+24/-24) src/libmatrix/mat.h (+17/-16) src/program.cc (+1/-1) |
To merge this branch: | bzr merge lp:~jammy-zhou/glcompbench/glcompbench |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Jesse Barker | Disapprove | ||
Alexandros Frantzis | Pending | ||
Review via email: mp+57419@code.launchpad.net |
To post a comment you must log in.
Unmerged revisions
- 14. By Jammy Zhou
-
The transpose parameter of glUniformMatrix4fv must be GL_FALSE for GLES2
If GL_TRUE is set, GL_ERROR_
INVALID_ VALUE will be returned. So use column
major order for matrix transformations.
As discussed previously, I'm concerned that the actual and advertised behavior of libmatrix in these changes are consistent (and that cases apart from the ones being exercised in the current state of glcompbench are satisfied). I've pushed a new version of libmatrix that provides column-major data to the API, but a row-major interface to the programmer (IMHO, much more intuitive). The result is that rebranching the instance of libmatrix in glcompbench would require only changing the transpose parameter to glUniformMatrix 4fv().
cheers,
jesse