launchpad-dev team mailing list archive

[Robert Collins <robertc@xxxxxxxxxxxxxxxxx>]

On Fri, Feb 11, 2011 at 4:48 PM, John Arbash Meinel
<john@xxxxxxxxxxxxxxxxx> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 2/10/2011 9:02 PM, Robert Collins wrote:
>> On Thu, Feb 10, 2011 at 10:59 AM, John Arbash Meinel
>> <john@xxxxxxxxxxxxxxxxx> wrote:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> I'm still getting rejection messages trying to post any merge-proposal
>>> stuff for loggerhead.
>>
>> I hope that that is fixed now.
>>
>>> I'm also stuck with stuff that was originally proposed against the old
>>> lp:loggerhead, which is no longer the "future" branch, and I don't have
>>> any idea what branch I'm actually supposed to be proposing against or
>>> merging into. (nor what the correct method for doing so is.)
>>
>> If you are finessing the future stuff that was put off into the
>> experimental branch, propose a merge to it and get it reviewed by a
>> loggerhead-team member (which includes all of ~launchpad - so lots of
>> folk that can help). To land in that branch, just land directly. I'd
>> like to suggest that only direct fixes to make it ok to land that
>> branch should be done there (that and merges from trunk to keep it
>> fresh).
>>
>
> Landing stuff directly to loggerhead/experimental seems reasonable. But
> I'd rather have PQM in front of lp:loggerhead (trunk). Is it reasonable
> to ask for that?

Totally, I prefer that too.

> I would assume we could use launchpad@xxxxxxxxxxxxxxxxx, but I'd sort of
> rather not. "bzr selftest -s bp.loggerhead" takes less than 10s, and it
> really sucks to have that queued up behind 4 Launchpad merge requests.
> (which seem to take ~30min even though they don't run the test suite.)
>
> For now, I can certainly be vigilant about running the test suite. It
> feels funny to have it finish so quickly.
>
> Though I'm also not testing against python-2.5, which I would like to
> preserve compatibility with.

We could get a pqm or a tarmac configured, for sure. If we just add a
rule to the existing bzr pqm that probably meets all your needs and
would be easiest. Would that suffice? We'd need to get a one time sync
of keys from the lp pqm to get lp folk commit access.

Alternatively, there *is* a simple tarmac for lp-oops and some other
things, I'm reasonably sure we can add loggerhead/trunk to that easily
too; I shall enquire next week when I overlap the relevant folk.

-Rob