[vdr] Re: dxr3/vdr problem

[Kai Moeller <moeller.ki@gmx.de>]

On Wednesday 05 March 2003 21:55, Marcus Metzler wrote:
>  > At the moment there are no information channels about defect packets in
>  > vdr (such channels are mentioned but not defined within standard).
>
> Since vdr is converting to a program stream it should also check for
> errors and if possible correct them. Unfortunately, this is not
> trivial and would mean quite some work. I don't think throwing away
> defect TS packets would help.

That's is more or less my opinion. I'm not sure whether VDR has to correct the 
errors or not. At least it would be nice to have the information about the 
possible defect ranges within the PES frames. What to do with these frames 
could be the task of the plugin. E.g.: The recording plugin could try to 
repair the stream with the help of a mpeg-lib/tool. For the dxr3-plugin I 
would decide to skip these frames and to stop the output if the error 
propability exceeds a certain level (repairing the stream may be to cpu 
intensive for small systems).
For the output on another mpeg-card it might be possible that the hardware of 
the card (or the firmware) provides such a functionality ...

>
>  > And for me it makes no sense to check the PES / ES stream  for potential
>  > bit errors if this knowledge was present in the lower layers (for a
>  > error free stream this would be only a waste of cpu time).
>
> In that case you would have to be sure that the information about the
> errors is correct. There could still be errors that were not detected.
>

Right. But the propability for undetected errors should be extremly small. As 
far as I know 16 bytes are used to protect the the 188 bytes of the TS 
packet. This is pretty much.

Anyway. I think, too, that this is a lot of work and I don't know whether this 
is really such a big problem (my system crashs only on osd problems ;-).
From a dxr3-plugin point of view I see at the moment no way to solve Malcom's 
problems (sorry).
If anybody finds bugs in our PES handling or quick checks (which don't consume 
much cpu power) don't hesitate to contact us (I'm working on a better buffer 
handling at the moment).
By the way, I'm using the lastest cvs dxr3 driver and for me it seems that 
this driver works more reliable than the official one.

Thank you very much for the information, Marcus.

- Kai - 






-- 
Info:
To unsubscribe send a mail to listar@linuxtv.org with "unsubscribe vdr" as subject.