[vlc-devel] commit: Remove non-sensical NPT computation code from c9569b35. ( Rémi Den is- Courmont )

Derk-Jan Hartman hartman at videolan.org
Fri Jul 25 01:25:57 CEST 2008


QTSS has to be broken.
I just added debug to liveMedia at the exact spot where the pts is  
read from the data and the following is shown:

[00000450] live555 demux debug: PTS 1216940140208000, diff for track  
mp4a, is now: 96000
timestamp 1007115067
[00000450] live555 demux debug: PTS 1216940140180969, diff for track  
avc1, is now: -33366
timestamp 1007115667
[00000450] live555 demux debug: PTS 1216940140281069, diff for track  
avc1, is now: 100100
timestamp 1007115467
[00000450] live555 demux debug: PTS 1216940132966634, diff for track  
avc1, is now: -7314435
timestamp 1956623927
[00000450] live555 demux debug: PTS 1216940140304000, diff for track  
mp4a, is now: 96000
timestamp 1007116067
[00000450] live555 demux debug: PTS 1216940133066734, diff for track  
avc1, is now: 100100
timestamp 1007115867
[00000450] live555 demux debug: PTS 1216940133033368, diff for track  
avc1, is now: -33366
timestamp 1007116467
[00000450] live555 demux debug: PTS 1216940133133468, diff for track  
avc1, is now: 100100
timestamp 1956626999
[00000450] live555 demux debug: PTS 1216940133096000, diff for track  
mp4a, is now: -7208000
timestamp 1007116267
[00000450] live555 demux debug: PTS 1216940133100102, diff for track  
avc1, is now: -33366
timestamp 1007116867
[00000450] live555 demux debug: PTS 1216940133200202, diff for track  
avc1, is now: 100100
timestamp 1007116667

So unless the data is read incorrectly from the packet, or if the  
packet is corrupt (both of which i seriously doubt), QTSS is not  
really sending the data as it should (can you confirm that Ross ? )

I've been considering how to handle this, and the only thing i can  
think about is ignoring a sample of which the Pts differs more a sec  
from the previous. ( 1000 000 ). (This is a common issue btw. I have  
seen this problem with most of Apple's keynote streams etc)

Apologies to Ross for implying that it could be liveMedia that was at  
fault, i just needed to cover all my bases in looking into this. It  
seems we are off to the QT mailinglist.

DJ



More information about the vlc-devel mailing list