 |
With some added debug code I was able to find the stacktrace where entity ref gets deleted It's still puzzling why the viewDef keeps a pointer to it, considering that they get rebuilt every frame image.png (28,800 bytes) |
|
|
Next stop in R_SetEntityDefViewEntity Enumerating each added viewEntity_t after the delete above only show valid entity ref's However RB_ShowViewEntitys receives a chain of viewEntity_t with an invalid pointer between entities 10 and 15 At this point I start suspecting the backend is not rendering the actually last frame data available, even with SMP off |
|
|
Assigning to @cabalistic as related to frameData management |
 |
You're right in that the backend was always one frame behind, independent of whether SMP is active or not. I changed that with 9527 and can't reproduce the crash anymore, so I'm hoping that's all there is to it. |
|
|
Can't repeat the crash any more on the current svn |
 |
I had to revert rev 9527 because it broke mission success screen. Instead of normal background and the skull, the last in-game frame is simply stuck on-screen when mission ends. Perhaps mission failure screen has the same problem --- did not check. Feel free to restore the original fix, but please fix and test mission success (use tdm_end_mission), mission failure, save/load, i.e. some typical switches =) |
 |
Too close to 2.10 release candidate. Moving to 2.11 |
- Anonymous
