bzr 2.0.2 ubuntu StableReleaseUpdate

[Martin Pool]

2009/12/14 Robert Collins <robertc at robertcollins.net>:
> On Mon, 2009-12-14 at 19:33 +1100, Martin Pool wrote:
>> OK, I put a proposed package onto
>> <https://bugs.edge.launchpad.net/ubuntu/+source/bzr/+bug/437626>.
>>
>> One thing we identified in feedback is: all changes must be justified
>> by a bug report with (if at all possible) a clear reproduction recipe
>> in the description so they can be tested.  Not all of our changes have
>> bugs associated with them.  It's not mandatory for trunk but we should
>> check all proposed merges to the stable branch have a bug report.
>> This isn't something anyone did wrong, just an improvement from here
>> on.
>
> Is this based on Martin Pitt's comments? I didn't read them quite the same way.
>
> Remember too that this is from an Ubuntu perspective - so simply having
> 'bugs' won't be sufficient - they need to be bugs 'in Ubuntu', if that
> is indeed the case.
>
> AIUI though, the key objectives from the Ubuntu side are:
>  - assess the risk of destabilising existing users
>  - check the problem is indeed fixed.
>
> For the former, I think our existing policy of only putting low risk
> changes into stable branches is sufficient, and shouldn't need changing.
> Adding a bug report as housekeeping isn't particularly efficient, and
> crucially - makes doing the SRU /harder/ because there is another case
> for the Ubuntu folk to check and sign off on.
>
> For the latter, having a test that we added or changed when we made the
> bug fix should be enough most of the time, but we can certainly consider
> having a small script.
>
> I think we should feel happy to talk with the Ubuntu devs in more detail
> about this: what we want to achieve using an SRU is not what SRU is
> designed to do - and so we should be precise about how things are
> fitting/not fitting and crucially why certain things are looked for.

OK.  I put up a doc mp
<https://code.edge.launchpad.net/~mbp/bzr/doc/+merge/16117> discussing
this.  Let's talk to them to be sure about that.

Ultimately finding out whether they accept it is the real test.  I
haven't seen any other feedback since my work yesterday.

-- 
Martin <http://launchpad.net/~mbp/>