On Jun 8, 2015, at 2:42 AM, Amber Hawkie Brown email@example.com wrote:
On 8 Jun 2015, at 17:27, firstname.lastname@example.org wrote:
On 05:57 am, email@example.com wrote:
As mentioned in http://twistedmatrix.com/pipermail/twisted- python/2015-March/029258.html , Twisted is removing Python 2.6 support. As such, I would like to announce that 15.3 will be the last release with 2.6 support. After this release, the Python 2.6 buildbots will be removed from the 'supported' list and most likely retired.
This will remove all buildbot coverage of OS X and RHEL/CentOS. As I've been taking point on the buildbots recently, I'm going to be deploying a new CentOS 7 buildslave which will provide coverage of that platform. That just leaves OS X uncovered -- so if you would like to donate buildslaves for this, please send me an email.
I think this is backwards. There was basically no strong argument made against dropping Python 2.6 support but there was no case made for dropping OS X support. I don't think you can remove the only OS X builder from the supported list. That's equivalent to saying OS X support is being dropped.
So either the existing OS X slaves need to be upgraded first (I don't see why you wouldn't just do this) or Python 2.6 support (at least on OS X) needs to be retained until someone else volunteers a Python 2.7 OS X slave.
I know that this would leave OS X uncovered -- lukasa, habnabit, and Adi have at separate points offered OS X builders of varying OS levels and even Python versions, both on the mailing list and in #twisted-dev (where Lukasa answered the call from this email and offered a Yosemite builder today), which is why I am confident that this will not end up being an issue when 15.3's first prerelease is issued.
I think the prevailing reason that we don't have an OS X 2.7 bot is that last time OS X hardware was offered, the buildbot was not being actively maintained. I've taken up the mantle in some capacity and getting Twisted supported on more platforms is my primary goal. And as an OS X user, I have some vested interest in it continuing being supported.
I think JP's point is that instead of saying "the OS X builder is going away, and if someone steps up there will be another one" we should say "we will make alternate plans for OS X before shutting off 2.6 support". I'm inclined to agree; if it's not going to be hard to do the upgrade anyway, we should make it a precondition.