Re: v0.56 released

[Dan Mick <dan.mick@inktank.com>]

On 01/02/2013 08:00 PM, Dennis Jacobfeuerborn wrote:
> On 01/02/2013 09:46 PM, Sage Weil wrote:
>> On Wed, 2 Jan 2013, Dennis Jacobfeuerborn wrote:
>>> On 01/02/2013 07:11 PM, Sage Weil wrote:
>>>> On Wed, 2 Jan 2013, S?bastien Han wrote:
>>>>> Debian-testing shows the version 0.56-1, maybe I misunderstood but I
>>>>> thought that 0.56-1 bobtail was the new version of the stable branch.
>>>>> So I was expecting to see it here
>>>>> http://ceph.com/debian/dists/precise/main/binary-amd64/Packages
>>>>>
>>>>> Correct me if I'm wrong :)
>>>>
>>>> It will be soon, but we're doing some additional validation before calling
>>>> it bobtail.  We may also move the URLs around a bit... more on that soon!
>>>
>>> What happens if you find a problem that needs a small update? Will the
>>> bobtail version be bumped to v0.57?
>>
>> v0.56.1, v0.56.2, etc., will follow in this series and bugs are found and
>> backported.
>
> That sound like an odd and potentially confusing versioning policy because
> v0.56 now is apparently a sort-of-stable-but-not-really version. I think if
> would be better to either go with something like v0.56rc1 until the release
> is actually declared stable or as some other projects do it use something
> like v0.55.99 as a pre-release version. That way you know if someone refers
> to v0.56 they actually mean the stable release.
> Right now if you find a problem and have to release v0.56.1 as bobtail then
> every time someone refers to v0.56 you have to ask them if they mean
> pre-bobtail v0.56 or post-bobtail v0.56. This kind of ambiguity should be
> avoided if possible.
>
> Regards,
>    Dennis

The versioning in argonaut went 0.48argonaut, 0.48.1argonaut, 
0.48.2argonaut, etc.  i.e. the codename is the series/branch of 
patch/fix releases, not just one point in the stream.  The 
truly-unambiguous version comes from any of the tools' -v option, which 
outputs the coded release name and the SHA1 of the commit.