186 replies to this topic

[littlesat]

Thanks Huevos!!! Looks good...

 

Is it possible for you to help us to limit the patch to only the 'stuff' that is really required for supporting JESS...So at least remove the coloring stuff for debugging... 

 

For real good testing one of our devs still need a JESS setup... so at this moment on short term we cannot test/verify/improve... So for a short term solution we need a good clean commit suggestion...

[arn354]

I think the suggested patch misses the last modifications from january till "today".

For example GuardTuningWord_a is not included in sec.h in the proposed patch.

https://github.com/o...054bc95f788132c

[littlesat]

fix timing problems with sendDiseqc and nonblocking ioctl …

 

 

This last one looks like a terrible patch...

 

I'm afraid the JESS application needs rework....

 

It gives me the "feeling" this is somehow required to solve driver issues.... (But I can make a mistake here)

Edited by littlesat, 4 March 2016 - 15:29.

[arn354]

 04-03-_2016_15-31-03.png   65.59KB   6 downloads

 

i think you just looked up dvb-adenin and missed my changes (which are also from dvb-adenin, but needed a change/correction).

[arn354]

it is maybe "partly" driver related - but it has more to do about some timings of unicable components/periphals which were causing issues.

I did not have an issue before that patch with my boxes and my unicable/jess setup - and i do not have an issue afterwards. But that patch solved the issue for others in openATV.

Edited by arn354, 4 March 2016 - 15:46.

[arn354]

i just asked adenin - yes seems drivers handle what is happening during sendtime differently. some lock e2 for about 90ms (14.5ms per byte). so it is basically a timing issue between e2 and the drivers from what i understood.

[littlesat]

Not nice these kind of work-a-round

[Huevos]

OK, I've started again from scratch. Added all the commits mentioned by Arn354 and the .xml file. And completely removed the "colours" code.

 

Right now it merges cleanly with the current PLi master branch, and builds fine.

 

https://github.com/H...7a4a2599ffd4d08

[Huevos]

If anyone in the PLi team would like test image with this code I have Solo4K and Solo2 built.

Solo2 is up and running with one cable and internal loopthrough. Record a channel and it is still possible to zap to another channel on any of the 4 sub bands.

I've been testing since yesterday and no strange behaviour has been encounted.

[littlesat]

Nobody of the pli team can test it at this moment as we do not own a jess system.... So at this moment for quick progress we can only verify the quality of the code and trust you by the colour(s) of your eyes...

As far I verified the code I see a huge patch...I have also and this is just a feeling the idea it could be made much less complicated... But for know it is the best we can get (quicky). As I/we cannot test we also cannot check if it could be simplified...

Edited by littlesat, 6 March 2016 - 09:20.

[Huevos]

Littlesat, I don't understand what you mean. You asked for a patch. Then you asked for the debug stuff gone. All that is done. It has been a lot of work to get to this stage.

[Huevos]

The only reason it is a big patch is because it is code that has been developed and perfected over the period of a year. Steady progress and testing lots of different pieces of equipment. This is mature code that didn't just appear yesterday.

[littlesat]

Sorry I do not want to trigger bad feelings.... We do appreciate your work very much,,, And I know it was a 'hell of a jub'.....

[littlesat]

I just created a jess branch that includes your suggested patch... 

 

https://github.com/O...igma2/tree/jess

https://github.com/O...a2/commits/jess

https://github.com/O...9d7a8933063cf01

 

I like the typo fix on top of the patch....structur...

Edited by littlesat, 6 March 2016 - 10:28.

[Taykun345]

I also like your typo "hell of a jub"

[Erik Slagter]

Littlesat, I think Huevos means that we should test the code if it doesn't break non-Jess installations, no regressions. Right?

[littlesat]

I have also no regressions.... and as far I can see here it does not break anything here.... So it "looks" good that way

 

But I made a branch for it....

 

Nice typo jUb... isn't it???

Edited by littlesat, 6 March 2016 - 11:25.

[Huevos]

Littlesat, I think Huevos means that we should test the code if it doesn't break non-Jess installations, no regressions. Right?

Yes, exactly. Code affect all hardware.

[Erik Slagter]

Littlesat how did you test? If you only recompile enigma, that's not enough, you need to replace the whole package for proper testing.

[blzr]

briefly tested on et9500 / vuzero which share same unicable(1) installation - definitely did not look good,

when e.g. set recording on tuner A (scr1), zapping to any other service (tuner B scr2) took over 5 seconds, then sometimes picture froze after few seconds, or couldn't get picture at all

 

switched back to master enigma for now...

 

will try to test more in the afternoon (when I'm back from ice rink )