Proposal discussion for Cantor: Python 3 as the only backend officially supported

[Albert Astals Cid]

El dimarts, 9 de gener de 2018, a les 14:41:54 CET, Aleix Pol va escriure:
> On Mon, Jan 8, 2018 at 7:10 PM, Filipe Saraiva <filipe at kde.org> wrote:
> > Hello gearheads,
> > 
> > Yesterday I wrote a post in my blog about a proposal to Cantor.
> > Currently, Cantor has 11 backends but most of them are not maintained.
> > The consequence is several backends are broken or not working as
> > expected. I waste a lot of time trying to fix backends I don't use, and
> > I don't have time to work in new features for Cantor itself.
> > 
> > My proposal is officially maintain only Python 3 backend, move the other
> > backends to a community/third-party repository and, if someone cares
> > about them, release some of them as extensions in KDE Store.
> > 
> > This way I can work better with Cantor.
> > 
> > Just to say, I tried in past solve it inviting developers to be
> > co-maintainers of the backends, but it does not work as I expected.
> > 
> > I would like to present it to KDE community and hear you.
> > 
> > The post with more information is in
> > http://blog.filipesaraiva.info/?p=1897
> 
> I'm not sure it's a great idea. If you want Cantor to be python-only
> that's one thing, but moving the maintainership elsewhere will
> probably make contributions just harder.

Kind of agreed, if you want to kill all the code without maintainer that's 
fine, but let's not pretend someone will take care of that code once you move 
it out.

It will be harder to take care than it is now, so i doubt anyone will.

> 
> Also note that we can't have compiled binaries distributed reliably
> under the store.kde.org.
> 
> My recommendation is to take QA more seriously by including proper
> tests on the several backends and running them instead of ditching
> everyone's contributions. But of course, if something isn't working,
> it should not be released.

The problem here is defining "isn't working", let's give an example of Okular, 
we have support for the xps format, it's not perfect, but it gives people 
something they may be able to use without rebooting into windows.

So without knowing anything about the status of the non-python3 backends, how 
useful are they?

Are they usable but have bugs?

Or are they totally broken?

If they are usable but have bugs, i would suggest to leave them in and when 
there is a bug, if you feel you don't really care enough, comment something 
along the lines of "This backend is unmaintained but it is still enabled 
because some people use it, if you want to help improve it, please get in 
contact". Optionally having a messagebox in the app saying something similar.

If they are totally broken, i would still suggest to leave the code in the 
app, but maybe hidden behind a compile switch, so it's easier to bring it back 
when it gets enough quality.

Best Regards,
  Albert

> 
> Aleix