[Python-Dev] death to 2.7; long live 2.7

[Ned Deily]

In article 
<CAP7+vJLVrqFHEFsHN+i3mRaHEtvZa_POh9OuLmhYAv6pBNWO4A at mail.gmail.com>,
 Guido van Rossum <guido at python.org> wrote:
> On Wed, Apr 9, 2014 at 9:58 PM, Benjamin Peterson <benjamin at python.org>wrote: 
> > It's not that I don't think Windows installers are important, but rather
> > that Martin has indicated he is (completely reasonably) not interested
> > in indefinitely making 2.7 installers. 
> Yeah, this was mentioned a few times. I quipped to Nick that Red Hat's
> biggest contribution might be to take over the Windows Installer, but he
> didn't bite. :-)
> 
> But there's always the PSF. We may try to find some folks we trust with
> relevant expertise to volunteer their time in return for a stipend from the
> PSF for this and some other unglamorous tasks.

WRT the other set of installers we provide, I'm willing to keep 
producing OS X installers for an indeterminate future of 2.7.  I reserve 
the right to get tired of it before 2038.  And I certainly am not 
volunteering to take over the Windows Installer.  I have it easy 
compared to Martin.

If we decide to keep going past 2015, I will likely propose some changes 
in the supported OS X levels of the 2.7 installers, in particular 
dropping at least 10.3 and 10.4 which we already did for Py3 starting 
with 3.3.0.  (By this I am not proposing to do anything to break source 
builds for those older systems.)  That has some potential impact to end 
users, e.g. breaking ABI compatibility in a minor release, but I think 
the impact would be outweighed by the benefits of supporting newer 
os-dependent features and build tools.  Any changes could be mitigated 
by a transitional release with installers for both the old and the new 
configurations.  But that's just a head's up: I'm not prepared to go 
into details at the moment.

-- 
 Ned Deily,
 nad at acm.org