[Python-apps-team] Bug#937009: Bug#937009: mercurial: Python2 removal in sid/bullseye

[Sandro Tosi]

> > > Do you need any help in coordinating with the packaged extensions,
> > > testing changes, preparing patches? a lot of time has passed since we
> > > started asking about mercurial and python3 and it is becoming the only
> > > reverse-dependency of several packages that could be removed if
> > > mercurial switched to py3k.
> > >
> > Getting an uptodate list of extensions and their status wrt porting both
> > upstream and in Debian would be useful.  I've spent some time looking at
> > hgsubversion a few weeks ago but there's a ton of work and I don't
> > actually use it so I've kind of given up on that; I forget what the
> > status is on others.
>
> will look into the rdeps of mercurial once 5.4.1-1+exp1 hits the archive

I've rebuilt all rdeps of mercurial against the python3 version
uploaded to unstable; results are:

2020/07/05 00:28:22 Build results:
2020/07/05 00:28:22 PASSED: salt
2020/07/05 00:28:22 PASSED: golang-github-masterminds-vcs-dev
2020/07/05 00:28:22 PASSED: pepper
2020/07/05 00:28:22 PASSED: python-hglib
2020/07/05 00:28:22 PASSED: git-remote-hg
2020/07/05 00:28:22 PASSED: haskell-filestore
2020/07/05 00:28:22 PASSED: composer
2020/07/05 00:28:22 PASSED: yotta
2020/07/05 00:28:22 PASSED: ros-rosinstall
2020/07/05 00:28:22 PASSED: check-manifest
2020/07/05 00:28:22 PASSED: jags
2020/07/05 00:28:22 PASSED: setuptools-scm
2020/07/05 00:28:22 PASSED: reposurgeon
2020/07/05 00:28:22 PASSED: devpi-common
2020/07/05 00:28:22 PASSED: firmware-microbit-micropython
2020/07/05 00:28:22 PASSED: python-hgapi
2020/07/05 00:28:22 PASSED: hgsubversion
2020/07/05 00:28:22 PASSED: ros-wstool
2020/07/05 00:28:22 PASSED: ros-vcstools
2020/07/05 00:28:22 FAILED: hg-git (see buildlogs/hg-git_0.8.12-1.2)
2020/07/05 00:28:22 FAILED: meson (see buildlogs/meson_0.54.3-1)

(build logs and artifacts are at
https://people.debian.org/~morph/mercurial_python3/ )

hg-git is RC and not in testing since mid-December, and meson is i
think a real error, since without mercurial depending on python2 now
the build fails to find that executable

Tbh, i think it's time to just rip the bandaid and upload mercurial
python3 to unstable, and deal with the consequences there (i volunteer
to do so); you mentioned hgsubversion as a potential blocker: that
package are popcon on 56, i dont believe such a minor package should
hold progress with the py2removal effort (I've added debian-python@ to
gather their input on this).

I do understand the rebuild results are not conclusive on the python3
migration (after all hgsubversion rebuilds fine with py3k mercurial,
which also raises the questions of why it b-d on it since it clearly
doesnt use it, but i'm digressing), but i think it's better to just go
ahead and upload the experimental version to unstable and see what's
the impact there

What do people think about this?

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