|Summary:||GL contexts are sometimes not displayed with XDarwin's AGL|
|Product:||xorg||Reporter:||Torrey T. Lyons <torrey>|
|Component:||Server/DDX/XQuartz||Assignee:||Torrey T. Lyons <torrey>|
|Status:||RESOLVED FIXED||QA Contact:|
|OS:||Mac OS X (All)|
|i915 platform:||i915 features:|
|Bug Depends on:|
Description Torrey T. Lyons 2004-08-27 11:31:48 UTC
With rootless XDarwin, when using the Cocoa implementation and the AGL GLX bundle for indirect accelerated GLX, some GL context are not displayed at all. The same client side code works properly with the CGL GLX bundle (direct or indirect) or the Mesa GLX bundle. An example is Tecplot <http:// www.tecplot.com/products/tecplot/tecplot_main.htm> whose main drawing view is always the grey window background color with AGL GLX. This essentially means that GLX is completely broken for some applications on Mac OS X 10.2 and earlier. The indirect code in Xserver/GL/apple for AGL (aglGlx.c) and CGL (indirect.c) are very similar. The AGL code should be fixed up following the example of indirect.c.
Comment 1 Kevin E. Martin 2004-08-29 12:23:58 UTC
We are very close to the release. Is this something that you could make a small self-contained patch for? If so, then I think we could include it. Otherwise, I think we might need to document it as a known issue for the release. I will bring this up on the release wranglers call on Monday.
Comment 2 Torrey T. Lyons 2004-08-29 14:56:41 UTC
I have basically finished a completely self contained patch to aglGlx.c which fixes this issue. It just needs some final clean up and testing. Assuming this works out okay, I'd like approval to put this in on Monday since this is not code that anyone else is familiar with. It is only used on Mac OS X 10.1-10.2 in rootless mode. However, the value of fixing this issue is very high to those users.
Comment 3 Kevin E. Martin 2004-08-29 20:28:20 UTC
If the patch is limited to a single file, then I think it will be fine. At this point everything is going through the release wranglers, so I'll bring it up on the call tomorrow and let you know.
Comment 4 Kevin E. Martin 2004-08-30 17:59:44 UTC
Unfortunately, we ran out of time before getting to this bug, but the release has been delayed by a day or two, so there is still time. I think that what you are proposing is fine. I have added Egbert to the CC list, so if he has no objections, then I would say that you should go ahead and check in your patch to aglGlx.c.
Comment 5 Kevin E. Martin 2004-08-31 18:08:12 UTC
Torrey, any status update on this one?
Comment 6 Torrey T. Lyons 2004-08-31 21:27:21 UTC
I had something come up today so I was not able to spend any time on this. I have a 90% correct patch that still has some artifacts in edge cases and I was hoping to get a 100% correct patch. I guess we should go with the best I have on Wednesday.
Comment 7 Kevin E. Martin 2004-09-01 10:33:50 UTC
(In reply to comment #6) > I guess we should go with the best I have on Wednesday. We discussed this on the release wranglers call today, and agree that whatever you have available today and feel is best for your platform should be what we go with. The deadline for all patches is today by 6pm EDT. Please attach the patch that you want to include and I will check it in. Thanks!
Comment 8 Torrey T. Lyons 2004-09-01 15:55:11 UTC
Created attachment 808 [details] [review] Partial fix for AGL display problems This fixes the most egregious problems. Unfortunately there is still an issue with flushing updates to the display after a resize. You can manually scrape a window (even non-X11 window) over the window after a resize to fix it. Oh well.
Comment 9 Kevin E. Martin 2004-09-01 18:38:36 UTC
I've checked in your fix and will close it since any further work will need to be done after this release. Another bug can be opened to block the next release after this release goes out.