[oi-dev] oi-build review process

[Guido Berhoerster]

Hello Alasdair,

* Alasdair Lumsden <alasdairrr at gmail.com> [2011-08-06 01:20]:
> Hi Guido,
> 
> On 08/ 5/11 11:42 PM, Guido Berhoerster wrote:
> >Hello,
> >
> >looking at the first review for oi-build I think it is suboptimal
> >that changes put up for review are already committed, changes
> >requested by reviewers can lead to a number of additional commits
> >which show up in the master repo and make changes more difficult
> >to follow, furthermore I would find it desirable to have
> >Reviewed-by headers in the commit message documenting our review
> >process. I'm not familiar with how Bitbucket works but off the
> >top of my head I can think of a number of solutions, changes
> >could be kept in MQ during review and then be turned into commits
> >before integration, there is a histedit Mercurial extension,
> >there is webrev and there may be further options.
> >
> >It would also be nice to define some minimum time period for
> >which stuff is put up for review in order to give interested
> >parties enough time to look at changes.
> >
> >Any thoughts/opinions on this?
> 
> I was expecting the committer to use "hg pull --no-commit" and pull
> each of the changesets from bitbucket, then commit and push to the
> master repo - that way there is only one commit per change. In the
> commit message it can reference the author + reviewers.
> 
> The beauty of using bitbucket is that it makes it really easy for
> new less seasoned developers to contribute, and using "--no-commit"
> solves the "messy commits" issue.

I suppose you mean that the contributor hg exports all changesets
and the committer then uses "hg import --no-commit"? If so, then
yes, that would do it. In that case we use something like
----8<----
<bug no> bug description
Author: Name <email>
Reviewed by: Name <email>
Reviewed by: Name <email>
[...]
---->8----
as the format for commit message?

-- 
Guido Berhoerster