[fixed] LiveTV - Zap some seconds back jumps to the Start of Timeshift. (1 Viewer)

[Brudertac]

Thanks for the recording

It may take a while to work out where the problem is (and fix it properly), but attached is a TsReader.ax with a workaround which should allow you to skip around (with lower accuracy than normal) in the 'problem' channels. Just unzip it and copy it into C:\Program Files (x86)\Team MediaPortal\MediaPortal (it is compatible with MP 1.2.3 and 1.3.0)

Tony

Awesome! It works very good now. I´m glad that you found something.
Is there a chance to see the final Fix in 1.3.0 Beta?

Many Thanks!!!!

Brudertac

 

[mm1352000]

Awesome! It works very good now. I´m glad that you found something.
Is there a chance to see the final Fix in 1.3.0 Beta?

I may have misinterpreted, but I think Tony was saying that he hasn't found the problem yet... and after that the problem still has to be fixed.

The TsReader that you're using is just a "hack".

@Owlsroost: it seems that pairs of PCR packets are being inserted by TsWriter. I'm still to confirm why that is happening, but I might know more in a couple of hours...

 

[mm1352000]

Aha, I see what is going on here... I think.
It is not TsWriter that is inserting PCR twice. Rather, it is that the PMT says that there is a dedicated PCR PID, but the video PID also has PCR timestamps with a different time base. By somewhat of a coincidence, TsWriter is rewriting the PCR into the video PID stream. This will obviously cause PCR timestamp clashes/confusion if the timestamps in the two streams aren't complimentary (ie. from the same clock)... which they aren't.

So really TsWriter needs to either use a separate PID for PCR, or it needs to wipe any PCR timestamps in the PID that it chooses for PCR.

The first solution is probably easier to implement and could consume less processor time, and in fact from the code it looks like that may have been how things were sometime in the past.

@Owlsroost, @tourettes, @miroslav22, @georgius, @DJBlu, @morpheus_xx, @misterd, @arion_p, @chemelli, @Ambass
Is there any real technical advantage, compatibility issue, or other reason for having the PCR in the same PID as the video?

Personally, I can't think of any technical advantage.
Compatibility with certain hardware or software could be an issue I guess.
Can't think of anything else...

 

[mm1352000]

Hmmm, maybe one other alternative would be to mark the adaption field with the OPCR (original PCR) flag instead of wiping it completely...

 

[mm1352000]

Fixing issues like this makes me happy.

Here is a potential fix that wipes the original PCR.
mantis #4034
github commit (code only - binaries can come later once the fix is confirmed)

The effective part of the fix starts at line 1111. The other part is just a juggle of the existing code so that I didn't have to copy-paste the code to a second location.

I chose not to go with the OPCR option as I don't think there is any need to keep the original timestamps, and it also requires pretty much the same amount of code and processing to ensure there is not already an existing OPCR that would have to be wiped to enable the PCR -> OPCR switch.

We could still go with the dedicated PCR PID option very easily. At this point I feel there must have been some kind of reason to put the PCR into the video stream so it seems safer to stick with that approach...

@Brudertac: if you want this fix to be included in MP 1.3b, please could you revert to the original TsReader and try the attached TsWriter. Note that it won't fix your existing RTL recordings - it only fixes the problem for timeshifting and future recordings.

@DJBlu: this fix might well resolve the similar issue with Setanta Sports 1 Ireland that you mentioned the other week...

 

[Brudertac]

Fixing issues like this makes me happy.

Here is a potential fix that wipes the original PCR.
mantis #4034
github commit (code only - binaries can come later once the fix is confirmed)

The effective part of the fix starts at line 1111. The other part is just a juggle of the existing code so that I didn't have to copy-paste the code to a second location.

@Brudertac: if you want this fix to be included in MP 1.3b, please could you revert to the original TsReader and try the attached TsWriter. Note that it won't fix your existing RTL recordings - it only fixes the problem for timeshifting and future recordings.

@mm1352000,

Glad to help to make you Happy.
With this Fix it works perfect! Can´t find any Error now.

Many Thanks to all who Helped so fast here!

Brudertac

 

[DJBlu]

I'll check it out and let you know.

Cheers.

edit:

Working here m8. Good spot. Thanks.

 

[mm1352000]

Thanks for the feedback guys.

So we know what the problem is.
We have a potential solution.
Are we happy with the approach and the code?
I note that this fix has the potential to do bad things to MP and TV Server if it isn't 100% perfect. I don't want to merge it without wider testing and/or a consensus that the code is okay. Alternatively, we merge it immediately to get as much internal testing as possible and keep an eye on reports in the beta.

 

[Owlsroost]

I'll take a look at the code later, but it sounds like the right way to fix it

Tony

 

[Owlsroost]

I'll take a look at the code later, but it sounds like the right way to fix it

Tony

Code looks OK to me

I think I'd go with merging it ASAP for internal testing - it's a self contained mod, so easy to roll back if we had to.

Tony