56 replies to this topic

[PainHealer]

Ok,

makes sense. But why does it wind reverse (even if not really clean) but stops when trying to wind forward? Is there anything that can be done against this issue? Not that I need default speeds higher 8x at all, but in reverse winding it always looks far away from clean and I feel like winding speeds over all are far away from "realtime".

I guess this behavior is mostly true for spooling in timeshift mode? The CT9100 is my first "e2" based STB, but at this moment I really don't like the way spooling is implemented at all. Is there any chance this will be fixed/improved in the future? Compared to most other "closed source" STBs it does looks "ugly" and is not somthing to be happy about in daily use. Does it depend on the used hardware or is just just a software topic?

Thanks for your help and the hard work on OpenPLi!

Regards

PainHealer

[PainHealer]

Ok,

one more addition to this topic: At Clarke Tech Forums I was asked not to "mix up" two different problems in this thread.

Just to clarify this: Besides the issue with the "default wind speed higher than 8x" there is still the other issue where certain recordings can not be spooled at 16x (or higher), even if the default winding speed is 2x-8x. As already mentioned by Biki3 at the beginning of this thread, it looks like the .ap-files have to do something with this issue. On certain channels/tansponders (RTL, VOX, etc.) the .ap seems to be corrupted during the recording, which causes the playback to "hang" when spooled at speeds above 8x.

Regards

PainHealer

[MiLo]

Do you have an example of a channel that has this problem? (And by channel, i mean something that I can tune to, "RTL" is to vague for that, so satellite and frequency would be good start) And does the recording have to be long, or is 10 minutes or so enough to reproduce it?

[MiLo]

And is it the .ap file which is faulty (how do you know?), or just "either the .sc or .ap" (because running the reconstruction fixes it, and that reconstructs both .ap and .sc files).

Another thing, if you could save the original corrupted .ap file and the reconstructed, correct one, and post or send it, that might help already. Those are pretty small files.

[Pike_Bishop]

Hi MiLo,

Another thing, if you could save the original corrupted .ap file and the reconstructed, correct one, and post or send it, that might help already. Those are pretty small files.

yes i will do this today or tomorrow.

we have the spooling prob with freeze higher than 16x at the followig channels;

Rtl
Rtl2
Super Rtl
Vox
NDR
phoenix

and maybe nor by other channels too.

it is enough to reproduce this problem with a recording with 10 minutes lenght.
it looks as it's only the .ts.ap file corrupt.
when we remove this .ts.ap file then the same recording with spooling higher than 16x is ok.
when we remove all meta files and let EMC make the new files to spool then EMC make only the ts.meta and the .ts.sc file
so that spooling is also correct because the .ts.ap file was not created again,
but in this two examples we have artefacts at spooling.

when we make for the problem records the spooling files new with the plugin reconstructapsc so this plugin creates the .ts.meta the .ts.sc
and also the .ts.ap file new, and then we can spooling without problems and without artefacts.

regards
Biki3

Edited by Biki3, 30 November 2011 - 20:38.

[PainHealer]

Hello MiLo,

one user at CT Forums found that deleting the .ap alone does the trick. After deleting this file, the recording that formerly could not be spooled above 8x will immediatly be "spoolable" at all speeds. Also, using "reconstruc" will always help.

Attached you will find a set of .ap/.sc-files, before and after reconstruct. before it "hangs" at 12x/16x when inital wind speed was 4x.

Channels on Astra 19.2E where the problem usually occurs:

RTL Germany SD Astra 1L H 1.089 27500 3/4 12188
Vox Germany SD Astra 1L H 1.089 27500 3/4 12188

and there are other channels also. If needed, I will write down more.

Regards

PainHealer

Attached Files

•  before_recon_error_at_12x.zip   186.16KB   2 downloads
•  after_recon_ok_at_all_speeds.zip   196.73KB   2 downloads

Edited by PainHealer, 30 November 2011 - 20:45.

[Pike_Bishop]

Hi MiLo,

Another thing, if you could save the original corrupted .ap file and the reconstructed, correct one, and post or send it, that might help already. Those are pretty small files.

ok here i post now the corrupt and the repaired .ts.ap file in the attachement.
these are the files for the spooling problem with freeze by spooling higher than 16x.

Thanks for your interest.

regards
Biki3

Edited by Biki3, 30 November 2011 - 21:04.

[Pike_Bishop]

Sorry i forgott the zip file.

regards
Biki3

Attached Files

•  ts.ap.zip   117.83KB   2 downloads

[MiLo]

The .ap files are identical in both cases.

PainHealer's .sc files are different though.

[MiLo]

A 10-minute recording from VOX indeed has the problem. It's the "ts.sc" file that appears to be incorrect, the .ap file is okay.

[Pike_Bishop]

Hi MiLo,

thanks for reply.

The .ap files are identical in both cases.

i think you mean the files of my attached .zip, but i don't understand it because i have maked
the repaired .ts.ap file with reconstructapsc and with this spooling higher than 16x is full ok.
with the original .ts.ap file however spooling higher than 16x is not correct, here it comes to the freeze.
i don't understand why this is so.

A 10-minute recording from VOX indeed has the problem.

can you make to this record new meta files with reconstructapsc, then you have a new .ts.ap file and spooling is allright,
so you can compare then itself the original.ts.ap file with the .ts.ap file by reconstructapsc.

It's the "ts.sc" file that appears to be incorrect, the .ap file is okay

hmm, i'am confused -> i thought as yet that the .ts.ap file is the problem.

regards
Biki3

[MiLo]

Removing either "ap" or "sc" file will stop enigma2 using both. I'm confused as to what the files actually do, it appears both are needed, if you remove one, the other one won't be used either. That explains why removing the .ap file fixes the problem - it runs okay because it doesn't use the .sc file then. Which makes me wonder what use they are anyway, it runs just fine without them...

[Pike_Bishop]

Hi MiLo,

you are right, when only removing the .ts.ap file then spooling works higher then 16x
but in this records i can see artefacts specific by spooling rewind.
in records for they i make new meta files with reconstructapsc i can't see artefacts by spooling and this records i can also spooling higher than 16x correct.

but it's really a confused problem this spooling problem and it is also more confusing that it is not on records from all channels.
have we a chance for a fix to this problem in the future ?

thanks for your interest to this thread and for your reply.

regards
Biki3

Edited by Biki3, 3 December 2011 - 15:57.

[MiLo]

This commit appears to improve things. At least, the .sc files that enigma2 creates now match the ones from the plugin on VOX recordings.
http://openpli.git.s...366714205fb667f
Had to write a Python program just to analyze the files...

For the techies: The difference in the "bad" and "good" sc files is that the "bad" ones just lack about 1% of the data records. I can't explain how that would cause that winding behaviour. After the change, the .sc file that enigma generates matches the one from the plugin, except for the last few data records. I guess that happens on all recordings.

[PainHealer]

Hello MiLo,

thanks for working on a fix for this issue! It seems that the basic problem with the recordings is now fixed. I did three recordings from the "problematic" channels (RTL Germany, VOX, RTL II) today and everything looks good when winding during playback, even above 12x/16x.

However, the other problems mentioned - winding hangs/slideshow when inital winding speed above 12x and heavy artefacts while winding back during playback of a channel/event that is still activly recording, which is basically the same "effect" as when winding back in timeshift mode - are still there. I think we have already discovered, that this happens out of other technical reasons.

Do you think there is a chance to improve/fix the other winding issues also?

Best regards

PainHealer

[Pike_Bishop]

Hi MiLo,

This commit appears to improve things. At least, the .sc files that enigma2 creates now match the ones from the plugin on VOX recordings.
http://openpli.git.s...366714205fb667f
Had to write a Python program just to analyze the files...

Thanks for your great work.

regards
Biki3

Edited by Biki3, 11 December 2011 - 20:15.

[MiLo]

... heavy artefacts while winding back during playback of a channel/event that is still activly recording, which is basically the same "effect"

This one should be solved now (committed a few days ago). If you play a recording that's still running, you can now smoothly rewind and wind it. Even if you play it on another box than the one that's recording. As a bonus, writing the .sc files now uses considerably less CPU and memory, and parsing them should also be a lot more efficient (though I doubt you'll be able to notice it in real life). Also missing ".ap" files no longer have any effect on spooling (deleting the .ap file seems to have no ill effect at all, I noticed...)
This only works for recordings, the timeshift is still just as horrible as it was. And the "enter at 12x" is also still unsolved.