--- Begin Message ---
- To: "Steve M. Robbins" <steve@sumost.ca>, Julien Cristau <jcristau@debian.org>
- Cc: 744171-done@bugs.debian.org
- Subject: Re: Bug#744171: transition: boost-defaults
- From: Emilio Pozuelo Monfort <pochu@debian.org>
- Date: Thu, 02 Oct 2014 10:15:33 +0200
- Message-id: <542D09A5.80806@debian.org>
- In-reply-to: <3055202.c1nVXWLD7W@riemann>
- References: <20140411021118.27189.98841.reportbug@localhost> <2835209.KAGx1pzgHy@riemann> <20140709065504.GD3236@betterave.cristau.org> <3055202.c1nVXWLD7W@riemann>
On 10/07/14 06:31, Steve M. Robbins wrote:
> On July 9, 2014 08:55:04 AM Julien Cristau wrote:
>> On Wed, Jul 9, 2014 at 01:39:38 -0500, Steve M. Robbins wrote:
>>> On July 9, 2014 08:15:27 AM Julien Cristau wrote:
>>>> On Wed, Jul 9, 2014 at 01:03:49 -0500, Steve M. Robbins wrote:
>>>>> That leaves two Boost versions 1.54 and 1.55, which made me realize
>>>>> that
>>>>> the transition tracker is too pessimistic. Right now 1.54 is
>>>>> considered
>>>>> "bad", but it shouldn't be.
>>>>
>>>> Why not? I thought the whole point was moving things from 1.54 to 1.55.
>>>
>>> No, I don't think so. In my view, the goal is to release with at most 2
>>> boost versions. The reason for keeping multiple versions is precisely to
>>> avoid having to do hard transitions [1] and boost-defaults was proposed
>>> [2] to keep the sourceful uploads to a minimum.
>>>
>>> This had been working well (in my view) since 2009. Somewhere along the
>>> line the release team started demanding boost-defaults use the transition
>>> tracker. I don't quite understand why. But if we're going to use a
>>> tracker, IMHO the transition to track is AWAY from the oldest boost
>>> (1.49) to the two newer ones.
>>
>> We removed 1.49 from testing months ago,
>
> Sure, but there remains an open bug to remove it from unstable.
>
>> and for at least the last two
>> releases we've shipped with just one boost version. What's changed?
>
> I don't think anything has changed. In my view, the goal is to release with
> at most two versions. So if we have just one, that's fine.
There was just one reverse dependency yesterday (gpsshogi) so I added a hint to
remove both that and boost1.54. We only have boost1.55 in testing now, so I
believe this is over.
Regards,
Emilio
--- End Message ---