Closed natural-harmonia-gropius closed 1 year ago
would break when cycling the vid
would break when cycling the vid
fixed
But... would be easier to be freezed when cycling the vid
But... would be easier to be freezed when cycling the vid
Should have same behavior as change video-rotate
with master branch.
Should have same behavior as change video-rotate with master branch.
Not here.
As I said above, this PR has nothing to do with freezing. All it does is restart the child process when it should be restarted.
The problem is in the logic of the spwan()
, which has not been touched yet. So there must be a freezing problem with video-rotate
.
No this issue in the current master. https://github.com/po5/thumbfast/commit/7b32934a72435d596b4aeeb87aa77f23650cbad3
You ignored my comment here https://github.com/po5/thumbfast/pull/25#issuecomment-1256965657 It's not the same problem.
Because set vid no
in the master did not do anything.
In your video on the black screen shows thumbnails, and this is what to fix.
TODO:
sync_changes()
watch_changes()
, only execute the last request.Sadly, debounce didn't fix the freezing of the video due to rotation, and I can't reproduce the freezing due to cycle vid
We could use mp.register_idle() and then queue up changes in watch_changes() that then get sent after all changes were gathered.
It is called too frequently.
Sadly, debounce didn't fix the freezing of the video due to rotation, and I can't reproduce the freezing due to cycle vid
~In my test, no matter whether the vid is switched or not, it is easier to freeze than the master branch.~
Ah, I forgot to test the latest push. Freezing does not occur after retesting.
But It will not generate thumbnails after cycle vid: set vid no
,set vid 1
.
I don't quite understand why this is so, Change vid has always worked for me. https://user-images.githubusercontent.com/50797982/194635139-aa3ce4ac-8165-49ba-a9a2-53f0e4e97aa6.mp4
And I think maybe it makes more sense to add debounce to spwan.
Change vid has always worked for me.
Just using a normal video to test.
Another issue happened here.
Just using a normal video to test.
Also works fine for me.
Another issue happened here.
I can see same error when I set rotate.
All the problems are due to the fact that the child processes and also the file system are unobservable asynchronous processes and we can only guess that it completes
@hooke007 Does it still freeze under your test? If not, then is ready for review.
Does it still freeze under your test?
Nopo. But the msg https://github.com/po5/thumbfast/pull/25#issuecomment-1272013284 is too noisy. Did a larger timeout would help?
the msg is too noisy
No idea until https://github.com/po5/thumbfast/issues/11#issuecomment-1295865590 is fixed.
Did a larger timeout would help?
I don't think so.
@hooke007 Can only reproduce when hovering the timeline during file change, and that issue is present on master.
Actually I'm on the uosc PR version, disregard that as I haven't tested with a script that uses the regular API.
I will look at this PR in detail tomorrow.
effective_w, effective_h = 0, 0
is removed, add disabled = not (w and h) or ...
for vid=0.
Freezing issue happened again with the latest commit...
Fix https://github.com/po5/thumbfast/issues/21