qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Aidan Skinner" <ai...@apache.org>
Subject Code review process (was Re: Code review agenda)
Date Mon, 04 Aug 2008 10:00:29 GMT
On Fri, Aug 1, 2008 at 2:03 PM, Aidan Skinner <aidan@apache.org> wrote:

> I'm not sure this is really working anymore, and I'd like to discuss
> alternatives.

So, the problem I have with the way we're doing stuff just now is that
we don't do it regularly enough, and we end up with multi-hour slogs.
The pain from this becomes almost self-reinforcing, and I don't think
we're getting as high an energy/payoff ratio as we could.

So, I'd like to propose moving to an asynchronous review model for
most changes, and keep the call going but talk about the larger, more
complicated patches in more detail and ignore the more trivial ones
there.

There are two general models I can see for doing this, and a couple of
different ways between each. I have a definite preference, but I'll
just enumerate them right now.

The 20k foot view is commit-then-review (CTR), and review-then-commit
(RTC). Commit-then-review is how we've been running things up until
now.

Orthogonol to that, there's the actual review process.

Previously we'd relied on mail to qpid-commits, which is quite easy to
ignore, only works for CTR and it's hard to verify that anything has
been reviewed, let alone that everything has been reviewed. The RTC
inverse of this, the good old fashioned ask for patch approval from
the list, is easier to tell that it's been reviewed and to verify that
things have been reviewed.

Having long slogs through the commits was easier to tell that some
things had been reviewed, but we did get into a bad habit of skipping
things and skimming things in the name of time.

We could use Jira (every change has a Jira, right?), comment on that
and use the voting feature to indicate that it's been reviewed. I
can't see anything along the lines of tags so adding a
'needs_review'/'review_ok' to the issue seems out (oh Jira, why aren't
you as good as Bugzilla?)

There's specialist tooling available, such as Review Board
(http://www.review-board.org/), which is perhaps better suited than
Jira to this, but would require infrastructure (I doubt we could get
infra@ to suppport this, but I could probably help out if the load
isn't too big)

Thoughts?

- Aidan
-- 
Apache Qpid - World Domination through Advanced Message Queueing
http://cwiki.apache.org/qpid
"Nine-tenths of wisdom consists in being wise in time." - Theodore Roosevelt

Mime
View raw message