[mythtvnz] 0.25 seektables recording lengths..

[David Moore]

On 21/05/12 20:09, David Moore wrote:
> On 21/05/12 18:59, Brett wrote:
>>
>>> Damn. It seems there are multiple issues related, I think, to timing.
>>> Here's what I'm seeing:
>>>
>>> - Duration error. Seems to be on TVNZ only. Interestingly duration is
>>> correct if viewed during recording.
>>> - Seeking back to the start of the recording causes a glitch which 
>>> takes
>>> a second or two to clear. This is on all recordings, old and new after
>>> 0.25 upgrade.
>>> - Audio/video sync drifts after a while. All recordings. Can be 
>>> fixed by
>>> pausing playback.
>>> - Occasional very brief stutter in playback. All recordings.
>>>
>>> My system is almost up to date with upgrades (myth and OS). Might be
>>> time to raise the problems with myth devs.
>>>
>> I never used 0.25.
>> I'm running 0.25+fixes from git on lucid ..dvb-t freeview
>> I don't see any recording duration errors on TVNZ, my duration errors
>> were only on old recordings&  fixed by mythcommflag --rebuild.
>> A/V sync is good.
>>
>> Seeking back to beginning does cause 2-3 second hiccup..
>>
>> Video playback is bad compared to 0.24+fixes, stuttering like cheap
>> digital handi-cam recording. 24p&  i60 recordings seem to play okay (to
>> me).
>> This comparison was made by running 0.24+fixes&  0.25+fixes on the same
>> install/database etc.
>>
>> I toyed with the idea of trying to reverse the, new to 0.25,p predictive
>> frame dropping/repeating/timing "feature" but can't understand how it
>> works.
>>
> To be more clear: I am also running 0.25+fixes from mythbuntu repo on 
> lucid. Also wrong duration is not consistent. Sometimes it is ok on 
> recent TVNZ recordings, sometimes not. Bizarre.
>
> Where do you set up the predictive frame dropping/repeating/timing 
> feature?
I believe TVNZ is at least part of the problem. I have extracted PCR 
values from several TV1 recordings and have found two with PCR 
discontinuities. The values suddenly reset to near zero and start 
incrementing again. These files have playback glitches at the point of 
the discontinuities. Also seems the discontinuity indicator bit is not 
set in the MPEG TS packet Adaptation Field. If I understand the purpose 
of this bit I believe it should be set to tell the decoder that a PCR 
discontinuity has occurred so playback sync timing can be handled 
properly. I'm rapidly losing respect for TVNZ technical people. I wonder 
what else they're not doing right.