Yep, its' alright. The pa main loop only needs to be locked during the flush and cork calls. It doesn't matter if m_corked
is written while it's locked or not as it is not accessed by the main…
I wouldn't be able to test this, as I don't have an easy access to Linux machines at this time.
The PR would definitely benefit from more explanation of what's going on, and what the fix…
@neXyon That is actually a good find!
Unfortunately, with CoreAudio the issue is only partially resolved: after the cancel the scene is advanced one frame forward. So if you start…
Thanks for the detailed explanation! With these your code changes are much easier to follow. You are right, it is much better to have Blender actually react fully to the synchronization callback…
Thanks for the upstream PRs! To make things simpler, I would like to continue the review/discussion of upstream changes there. As soon as the PRs are merged upstream we can commit those changes to…
The audaspace part is fine I guess. I still suggest considering uping the minimum required ffmpeg version to get rid of old code instead of adding more ifdefs. Also note that there are these two PRs: #121947 and #121960.
Ok, I figured out that you mean pressing the Escape key which cancels playback running screen_animation_cancel_exec
. It appears that the first issue is at play here: AUD_seekSynchronizer
is…
I have never had that behavior, Blender always stops right where it is, when I pause playback (with Spacebar or the Pause button in the Timeline). Can you tell me how you trigger the behavior that…
@neXyon I should have clarified a few things, sorry:
- CoreAudio. The issue here was that the CoreAudio backend used (by mistake) the same code paths as JACK on the Blender side of…
Have a nice vacation then! :)
Ok, if you think that's the way to go, I don't have any good reasons to object. All good!
I have a hard time reviewing this, since I don't even know which problem this PR is supposed to solve? The title speaks about CoreAudio and then there are lots of changes in Jack and none in…