On Mon, Apr 09, 2018 at 12:55:04AM -0500, Jason L Tibbitts III wrote:
> >>>>> "ZJ" == Zbigniew Jędrzejewski-Szmek <zbys...@in.waw.pl> writes:
> ZJ> Please don't. This is a repeat of the original idea of having
> ZJ> separate python3 packages back when python3 was being
> ZJ> introduced.
> It seems that you are suggesting that pointless bureaucracy be kept in
> place purely because it slows down a process which you don't personally
> like.  If so, that's awfully passive-aggressive.
> If someone does wish too create a separate python2 package via a package
> review, will you also attempt to obstruct that review?

This is a misunderstanding. I wasn't speaking against the proposal
to have a fast-track review process. I was speaking against the proposal
to make those separate packages at all.

(If people create a separate package for python2 after all, I'm fine
with an expedited review.)

> ZJ> This was always a huge pain and waste of maintainer time.
> I'm somewhat confused; you seem to wish to make it take more time, not
> less.
> ZJ> Doing this on any massive scale would means hundreds (up to 2800?)
> ZJ> "new" packages, a way to burn massive amounts of maintainer time.
> Has anyone at all suggested doing this on a massive scale?  I certainly
> didn't.  I only suggested making it easier to handle the case where a
> package maintainer simply doesn't want the python2 subpackage to be
> generated from the main package.  I get the impression you took my
> suggestion and for whatever reason turned it into something else
> entirely.
> ZJ> Let's do this instead. We need more co-maintainership and more
> ZJ> co-operation in Fedora.
> But it has all of the problems I outlined.
> ZJ> Keeping an exisiting python2 subpackage is really no big deal.
> Perhaps for you.  Perhaps not for others.  If it was no big deal in all
> cases then why have any python2 packages been dropped at all?

That's a good question. So far I have seen the following:
1. dropping python2 support in leaf packages gradually and fix any
   issues one-by-one is easier then doing it in one step and then
   trying to rebuild everything.
2. dropping python2 support gradually makes people notice and raises
   awareness that they need to wean off python2. This applies to
   maintainers and non-maintainers alike.
3. removing support for python2 reduces maintenance burden.
4. apparently people don't like python2 and don't want to deal with it.

I share the sentiment behind all four points and I think they are all
valid. The problem of when to remove support is to a large extent a
matter of pushing work around. Removing support for python2 from a
package makes life a bit easier for the maintainers of that package,
and a bit harder for maintainers of dependent packages, and users. In
my experience, the work to keep support for python2 at an existing
level is not high, so when looking at the level of the whole distro,
the positive gain from dropping support is smaller than the negative
gain for people who have to update for that lost support. Thus, my
feeling is that right now, it's better to keep it. If something major
happens, like package upstream stopping support for python2, or
dependencies going away, then there's a good reason to stop supporting

devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Reply via email to