371 replies to this topic

[littlesat]

Let’s invite the owif people here… as this seems also being a factor here

[WanWizard]

It was just an example, the same is true for every other plugin.

 

The point is the remark works both ways: you can say "we can't upgrade to Py3 because some plugin won't work anymore", or "we can't stay on Py2 because some plugin won't work anymore".

 

My personal view is that actively maintained and used plugins (of which OWIF is a prime exampe) will move to Py3 because images will, and non-maintained plugins will stay at Py2.

 

Which of these plugins are more important to us and to our users?

[40H3X]

Precisely, we'll have to move forward.

[littlesat]

The question is what is the move forward….?

[WanWizard]

Not sure that is really a question.

 

If we don't move forward and phase out Py2, we'll end up either having to maintain all plugins ourselfs (as they will move on), or we will no longer exist due to lack of functionality (and possibly no team members left).

[40H3X]

Precisely, we'll have to move forward.

I ment try to move over to Py3 and what stays behind stays behind.

[Pr2]

I totally agree you should move forward, people are still use old no longer maintained plugins or skins. I read so many times that you cannot develop a proper solution because it will break plugin or some skins.
Switching to Python 3 is the good time to get rid of all of this old stuff.
Plugin developers or skin creators should also be ready to adapt there code, otherwise you cannot go forward.

[littlesat]

First we need manpower, mood, time and passion to fix this together… And when we have devs that can and are willing to fix this the second smart step is to freeze further development… I speculate this is a process that might take up to 3-6 months… and then indeed accept old stuff does stop working while new free stuff will work for free later …. Then first go for python3 and when this is completed the second step is update oe…. So a kind of roadmap…

It is a big project so most important is concentrate on get it working… don’t go for improvements or other changes in e2 (except for put it on the todo list).

This does also mean that improve offline decoding should be on that todo list… as making bigger changes to e2 is difficult while transferring to python3…

during transition it is helpful to only accept bugfixes that do not interfere with the py3 process…

Edited by littlesat, 25 September 2021 - 20:12.

[A.A.]

My opinion is that with the actual resources/developers most of the work should be done on enigma2.

All other deviations from OE can be postponed.

 

My 2 cents

A.A.

[WanWizard]

+1.

[littlesat]

+1….

[MCelliotG]

Plugin developers or skin creators should also be ready to adapt there code, otherwise you cannot go forward.

I already have!!!
 

[betacentauri]

I have started a new branch python3.

All commits from me until now should be compatible with Python2. As they are compatible I would like to merge them into develop branch in near future. If we wait weeks/months with the merge e.g. new "print 'abc' " are added in develop which have to be converted again.

 

 

Important:

I have used futurize (similar to 2to3). Not all changes are compatible with OpenPLi. E.g. the cmp fix adds

from past.builtins import cmp

We don't have that module.

Or the fix zip does this:

-                               list=zip(selection[1].description, selection[1].choices),
+                               list=list(zip(selection[1].description, selection[1].choices)),

As the use list as variable name, the fix won't work as far as I know.
So be careful with such automatic tools. They are nice, but not all changes will work!!

[littlesat]

List as variable name should be forbidden… I suggest to modify them al to list_description of listDescription or so… so also an indication of the content is given….

[betacentauri]

Yes, would be the preferred solution, but I fear list is used “everywhere”.
There are also other solutions to fix it.

I try to continue adapting the code in the next days/weeks, but I don’t know how much time I will have.

[littlesat]

Thanks….

[40H3X]

Nice we move forward

[dreamce]

Yes Openli 9.0 python3

[WanWizard]

The first reported issue: https://forums.openp...13#entry1392874

 

Python-pycurl doesn't support python2 anymore.

[Huevos]

-                               list=zip(selection[1].description, selection[1].choices),
+                               list=[x for x in zip(selection[1].description, selection[1].choices)],

It would be a nightmare if you changed those "list" variables as they are used in function keyword arguments by plugins, etc. Just use list comprehension instead. It shouldn't have been allowed in the first place.

 

For me option 1 is correct. Most mainstream plugins are already converted, e.g. OpenWebIF and I don't suppose Pli users have even noticed. All plugins at oe-alliance are converted and compatible with Py2 and Py3.

 

At OpenViX we have had a Py3 image for more than a year, with a limited public release of 6 months. The majority of the work was done by one team member on his laptop in hospital where he was having a knee operation. Then the rest was just polishing it all up.

 

From the user point of view they wouldn't have any idea what Python version is being used. It is just a nice stable image.

 

BTW, it is not just converting *.py code. CPP code has to be made compatible with Py3 in places.

 

Also, right now forget about end-of-line... the factories are still producing 100s of thousands of units every year.

 

https://github.com/O...hon3-compatible

Attached Files

•  1.jpg   99.21KB   15 downloads

Edited by Huevos, 15 October 2021 - 22:44.