[mythtv-users] PVR-150/300/500 HVR-1600 commflag workaround

[faginbagin]

Hoi Hika,

On 11/9/2014 12:35 AM, Hika van den Hoven wrote:
> Hoi faginbagin,
> 
> Sunday, November 9, 2014, 6:03:09 AM, you wrote:
> 
>> On 11/7/2014 10:36 AM, Carl Hunter wrote:
>>> On 11/3/2014 11:27 PM, faginbagin wrote:
>>>>
>>>> Thanks for testing the patch. I'm glad it, and the workaround, seem to have helped you.
>>>>
>>>> Regards,
>>>> Helen
>>>
>>> I finally looked into the typecast change and realized it was only re-compiling libmythtv-0.27.so.  I copied that and the new mythcommflag over to my live system and everything is working great.  My SD recordings now flag commercials with the typecast patch and mythcommflag does a much better job of marking commercials with your patches for SD content.  I would never get percentage lines from the following log snippet on SD content but now I do.
>>>
>>> I CoreContext ClassicCommDetector.cpp:367 (go) Finding Logo
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 0% Completed @ 97.3141 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 10% Completed @ 96.4692 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 20% Completed @ 96.7543 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 30% Completed @ 97.1697 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 40% Completed @ 97.2806 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 50% Completed @ 97.4146 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 60% Completed @ 97.5226 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 70% Completed @ 97.556 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 80% Completed @ 97.6085 fps.
>>> I CoreContext ClassicCommDetector.cpp:545 (go) 90% Completed @ 97.6055 fps.
>>> E Decoder avformatdecoder.cpp:4941 (GetFrame) decoding error#012#011#011#011eno: Unknown errror 541478725 (541478725)
>>> N CoreContext main.cpp:948 (FlagCommercials) Finished, 3 break(s) found.
>>>
>>> Now we just need to get your patches pushed to myth-0.27-fixes.  Any idea how to move this along?  Is there anything else I can do to test?
>>>
>>> Thanks
>>>
>>> Carl
> 
>> I wonder if it is appropriate to add comments to tickets saying "I
>> applied this patch, and it works for me". Maybe if enough people do
>> it, the devs will take the hint? I know they don't like "Me, too"
>> comments on tickets with no patches, but maybe "Works for me" on
>> tickets with patches would be received positively?
> 
>> Regards,
>> Helen
>> _______________________________________________
> 
> I have been checking on my corrupted seektables with my two PVR500's
> Helas, your bugfix doesn't solve that, but it made me wonder if a
> similar bug else causes it. Before 0.27 it was no issue for, as I
> understood on my mentioning here of the problem it was not used that
> way. It however confuses me still that it goes wrong on the run, but goes
> good on a rebuild afterwards. Obviously different codebases are used on
> the two different situations. Helas, I'm not versed enough yet to find
> such a spot.

You could be right. Maybe there's some difference between recording NTSC vs PAL using the PVR tuners? I don't really know the code well enough to say.

You say you didn't have the problem before 0.27. I believe the bug I fixed was there in 0.26. Did you use that version, or did you skip that release before 0.27?

I believe the bug I fixed only existed when making recordings on PVRs installed on 32 bit OSs. Are your PVR-500s installed on 32 bit systems?

The bug fix will only affect recordings made after the patch is applied, compiled, installed and mythbackend has been restarted. It won't help older recordings, for those you do need to rebuild the seek table.

> To explicate, on recording it looks like I get a speed shifted
> seektable, which shows on both jumping on playback and on the
> commercial detection.
> That brings me to a practical question. I tried encapsulating
> commercial detection in a script, where I first rebuild before
> detecting. It however is unclear to me, what is by default passed on.
> I tried:
>    #!/bin/bash
>    declare FileName=${1:-""}
>    [ "$FileName" == "" ] && exit 0
>    /usr/bin/mythcommflag --rebuild -f $FileName
>    /usr/bin/mythcommflag $@
>    exit $?
> Which does the rebuild, but no commercisl detection?

I assume you're passing only one argument to the script, the filename? Or, the filename is at least the first argument? It should be the full path name, not the basename. And you need to insert a -f in-between mythcommflag and $@. If mythcommflag doesn't like the arguments you give it, it will print a usage message and exit with a non-zero exit code.

Regards,
Helen