When should I drop python2 support for python-linecache2 & python-traceback2 (therefore, unittest2, mock, sphinx, pytest... and the rest of the Python 2 world if I pull this string until end...)

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

When should I drop python2 support for python-linecache2 & python-traceback2 (therefore, unittest2, mock, sphinx, pytest... and the rest of the Python 2 world if I pull this string until end...)

Thomas Goirand-3
Hi,

python-fixture (the binary) is gone, therefore we have #952130 and
#952127. I've been reluctant to remove Py2 support from these, because
unittest2 needs it, and this would break a lot of packages (including,
indirectly, stuff like sphinx, pytest, etc.).

What is it that the team suggests at this point? Should I reintroduce
Py2 support in fixtures, or should we go ahead and do a missive Py2 RM
of what's left in the archive?

We're down to only 366 packages with Py2 in testing. We can either
postpone forever, or just go hardly on it (but there will be inevitable
collaterals...).

Your thoughts?

Cheers,

Thomas Goirand (zigo)

Reply | Threaded
Open this post in threaded view
|

Re: When should I drop python2 support for python-linecache2 & python-traceback2 (therefore, unittest2, mock, sphinx, pytest... and the rest of the Python 2 world if I pull this string until end...)

Sandro Tosi-6
> python-fixture (the binary) is gone, therefore we have #952130 and
> #952127. I've been reluctant to remove Py2 support from these, because
> unittest2 needs it, and this would break a lot of packages (including,
> indirectly, stuff like sphinx, pytest, etc.).

you can remove them when

http://sandrotosi.me/debian/py2removal/python-traceback2_1.svg
http://sandrotosi.me/debian/py2removal/python-linecache2_1.svg

show no nodes with a red border.

> What is it that the team suggests at this point? Should I reintroduce
> Py2 support in fixtures

yes, it should have not been dropped since it still had reverse dependencies.

> or should we go ahead and do a missive Py2 RM
> of what's left in the archive?

that's just too much breakage.

> We're down to only 366 packages with Py2 in testing. We can either
> postpone forever, or just go hardly on it (but there will be inevitable
> collaterals...).

bullseye freeze is 9 months away
(https://lists.debian.org/debian-devel-announce/2020/03/msg00002.html)
so it's not acceptable to break reverse dependencies recklessly at
this time. let's take the slow and careful road, and work from leaf
packages backwards, as it has happened since now. we can re-evaluate
when we're closer to the freeze.

Regards,
--
Sandro "morph" Tosi
My website: http://sandrotosi.me/
Me at Debian: http://wiki.debian.org/SandroTosi
Twitter: https://twitter.com/sandrotosi