On Sun, 2020-01-12 at 17:07 -0500, Joshua Kinard wrote:
> I'm late to the party as usual.  Seems upstream plans a final 2.7.18
> security update in April of 2020, then they will consider the 2.7 branch
> EOL.  They say most of these updates were done in 2019, and so are still
> technically sticking to their mantra of no more updates after 01/01/2020.
> 
> PEP 373 covers this:
> https://www.python.org/dev/peps/pep-0373/#release-schedule
> 
> """
> Planned future release dates:
> 
>     2.7.18 code freeze January, 2020
>     2.7.18 release candidate early April, 2020
>     2.7.18 mid-April, 2020
> """
> 
> IMHO, I think we should retain 2.7.x compatibility for 1 year AFTER the
> release of 2.7.18.  This provides some time for people to transition systems
> off of 2.7-dependent packages.
> 
> It might be worthwhile to treat the removal of Python-2.7 from the tree in
> the same manner as an EAPI deprecation and removal, given how ingrained it
> is due to its longevity.  That will minimize the whiplash-effect of emerge
> complaining about slot conflicts and dependency conflicts.  Like I just ran
> into w/ setuptools-45.0.0.0's release.

Joshua,

I understand that you don't do much Python ebuild work, or probably
Python development in general.  I understand that you may feel like you
need more time with Python 2.  But before sending such mails, please try
to put yourself in our skin.

Imagine I've spend a few hours yesterday merely trying to find
a reasonable subset of OpenStack packages that can have Python 2 removed
(OpenStack has done Python 3 releases for quite a while already). 
Believe me, it's not interesting satisfactory work.  It's a struggle
against neverending conflicts with revdeps, stale stable packages, more
indirect revdeps...  All that to move a few dozen packages forward
and have less pain for end users in the end.

Now imagine someone who doesn't really know much about maintaining
Python in Gentoo and problems related to Python 2 sunrising, grabs one
site about Python releases and tells me what to do without knowing
the wider context.  Wouldn't you feel angry?  Demotivated?  Depressed
even?

I mean, forgive my expression but we're deep in shit.  As you've noticed
yourself, emerge spews few pages of 'I can't upgrade setuptools' because
of humongous number of packages that still need Python 2-capable
version.  Sure, we could put some effort into making it still work with
Python 2, then start collecting more and more patches to various
packages just to keep things working.  But then, 3-6-12 months from now
it will no longer be feasible, the cesspool will overflow and we'll be
even deeper in shit that we're today.

If people started removing Python 2 from Gentoo years ago, like upstream
suggested, today things would be much better.  But we waited till last
minute.  And now you're telling us to wait more because there will be
a new release of the *interpreter*?

-- 
Best regards,
Michał Górny

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to