371 replies to this topic

[A.A.]

I start this topic because I'd like to hear all developers voices about the transition towards python3.

 

Talking about OE/Yocto as of today meta-python2 layer is updated up to honister 3.4 (the next OE release, after hardknott 3.3) so enigma2 can be built and can run.

 

As for other recipes like samba, these can use python2 in hardknott but not in honister where the version is updated.

 

Kodi recipe is needing an upgrade to 18>19, this will be done soon after last finishes. We'll need a meta-kodi layer until the recipe flows in OE (if).

 

My feelings at the moment are mixed: I see many commits referring to the OE-A repository, where most commits are debatable.

 

I honestly would prefer to follow OE as much as possible and add less .bbappend(s) as possible.

 

A.A.

[WanWizard]

I'm not a developer, but my 2 cents:

 

The options are imho:

1. Upgrade E2 and our plugins to Python3 for OpenPLi 9 or 10.
2. Live with a dual Python 2 and Python 3 install, and forget about all boxes with <= 256MB flash.
3. Never upgrade to a newer OE again to avoid 2.

For me, there is only one valid option, 1.

[betacentauri]

Option 1 is much work, but it’s possible as we can see (e.g. ATV 7.0). I only see a lack of manpower. If only 1-2 developer adapt the sources after normal work, it will take months.

Option 2: Nothing what I prefer.

Option 3: Not upgrading OE also doesn’t make much sense. Earlier or later we need to. Or we need many workarounds to get everything running.

[WanWizard]

Option 4: None of the above?

 

We'll have to do something.

[Pr2]

Option 5: start something totally new. 

[ims]

ad 1) Is it possible use something as this ? https://docs.python....brary/2to3.html

[betacentauri]

ad 1) Is it possible use something as this ? https://docs.python....brary/2to3.html

Yes, sure. That can e.g. convert all print xyz into print(xyz) and other things. But you still have to do many manual work.

[littlesat]

Adapting e2 is one thing… but together with this all the (3rd party) plugins will become invalidated…. I saw openatv already converted some plugins… and note it is not only change the print stuff…. Betacentauri did years ago a part of the job…. And sorry I do not get the time….
2nd thing is that enigma2 is almost end of life… so should we invest at a moment teams do not work together….
I had always the feeling to go for the hybrid solution. But Ineas not aware yet that it needs such more flash. When we decide not to translate with hybrid we could decide only support the newer boxes and leave the older ones with non sufficiant flash on Zeus.

most optimal solution is indeed to get all converted….

Edited by littlesat, 23 September 2021 - 07:07.

[Dimitrij]

 

2nd thing is that enigma2 is almost end of life… so should we invest at a moment teams do not work together….

+1

[WanWizard]

Option 5: start something totally new. 

 

Which is what I've been advocating for years...

[40H3X]

I'm not a developer, but my 2 cents:

 

The options are imho:

1. Upgrade E2 and our plugins to Python3 for OpenPLi 9 or 10.
2. Live with a dual Python 2 and Python 3 install, and forget about all boxes with <= 256MB flash.
3. Never upgrade to a newer OE again to avoid 2.

For me, there is only one valid option, 1.

 

IMHO option 1 is the way to go, as it is the most future proof. Yet, when the future of Enigma2 is EOL, all the effort needed to achieve option 1 may not be worth your while.
 

[WanWizard]

IMHO there is a difference between "Enigma2 is EOL", and "our STB's and open software is EOL".

 

It would require a discussion about what the future may hold, in what direction we thing it will move, what kind of platform we envision, what kind of architecture and design that would need, etc.

 

Only after that we can decide whether or not option 1 is viable.

[littlesat]

Better solve ci+’issues and arrange offline decoding to extend lifetime and keep on older oe… and jump into something new? Maybe add iptv m3u support in a smooth way…

[WanWizard]

Thought you would say that. You've always been against modernisation, which is more and more blocking any progress.

[ims]

Sorry, it wasn't nice ...

[WanWizard]

Sorry, it wasn't nice ...

?

[40H3X]

 

IMHO there is a difference between "Enigma2 is EOL", and "our STB's and open software is EOL".

 

It would require a discussion about what the future may hold, in what direction we thing it will move, what kind of platform we envision, what kind of architecture and design that would need, etc.

 

Only after that we can decide whether or not option 1 is viable.

True, so let’s discuss. If I may kickoff, we know what providers want; “EOL open software”.

So maybe see what happens when “the die is cast” in feb 2022 (albeit a Dutch problem) and then make a decision?

[WanWizard]

As long as there are CI's, you can watch TV and record. And CI's won't disappear as most of them are used directly in TV's. Not a "dutch" problem, the traditional cardshare is approaching EOL.

 

For cable providers, the future is already becoming clear: Liberty Global has been introducing new boxes in several countries that don't have a card at all. They run a sort of softcam, which connects over the required internet connection to a "cardserver" at the provider. For satellite providers, that is more difficult to achieve, as an internet connection can't be required.

 

Cable and Satellite will continue to migratie to internet streaming, and view-on-demand instead of live TV, and at some point both well seize to exist, but IMHO that is quite a bit in the future.

 

Streaming (i.e. DRM) is already possible on linux, recording of DRM streams not, but since content remains available online, recording won't be needed any longer.

[rantanplan]

I would say that you are going very far away from the actual topic.

Why do you do it again and again.
I'm just a hobbyist and unfortunately I have very little knowledge of python.
I don't care if it's Python2 or 3.
You shouldn't talk things up as EOL beforehand.
Just keep tinkering and let those who have fun doing it.
If the python coders can do anything with python3, why not?
As long as the python2 image is still standing, the environment should be maintained for that long.
Maintaining means keeping up to date.

Bring some old good friends back on board and as much crossover as possible to the slightly different approach on the part of OE-A is perfect.
Crossover to neutrino would also be desirable.
Just don't piss off these closed source thieves.
That was a very bitter lesson for me too.

 

(sorry verry fast google translate)

[WanWizard]

I also don't care if it is python2 or python3. or whatever is used.

 

I do care about an "explosion" in the size of the image, and I do care about being restricted when it comes to migrating to new OE versions.

 

If the consensus will be "we're not going to deal with the EOL of Python2", the consequence will be that more and more development will become unusable for us (what if OWIF moves to Py3 for example?), and we'll be stuck on the last OE version supporting Py2.

 

I'm not an E2 programmer, so I don't have to do the work. But I personally do have an issue with the consequences.