On Tue, Feb 3, 2009 at 5:26 AM, Dean McNamee <[email protected]> wrote:

>
> Ever since I've been using git, I have super great and quick access to
> our full history and commit logs.  Being in a different time zone, I
> usually warm up in the morning by reading yesterday's commits.  I've
> found in general we are very good and strict about review code, but
> hardly anyone ever comments on the review description.  I often have
> to follow the code review link in the commit, and read through 30
> review messages to get the information I wanted.  A few problems I see
> very commonly:
>
> - Simple style issues:
> Typos / spelling errors.  Wacky line breaks, this is partially
> rietveld's fault, but we should standardize on something and fix
> rietveld / gcl.
>
> - Bigger issues
> A review description that should be a review message.
> Commit 1234:
> "Make this change.  Please review carefully, what do you think about
> the last half?"
>
> A "funny" commit message.  I'm not all dry, I like jokes, but they're
> not that funny in 1 year when you're trying to find a regression.
>
> A review description that is completely stale from the code.  When we
> iterate on rounds of feedback, we should also comment on the commit
> description to make sure it's updated.  Because of the way GCL works,
> we do the commit description first, and not last.  There can be a huge
> shift in the code between those points, and often the commit message
> goes forgotten.
>
> So, here is my proposal.  We review commit descriptions along with
> code reviews.  In general, we should think of every commit description
> as if it were a comment in the code.  That means proper spelling,
> formatting, up to date, etc.  It's important to remember that although
> you might be making a quick fix that is obvious to everyone in the
> room (hey, the tree was red!), in 2 days no one will remember that
> situation.  The code description will be left as 'qucik fix!', when it
> instead should have been "Correct a typo in XXX, the previous code
> fails to compile."
>
> Are you with me?
>
> Thanks
> -- dean
>


Great suggestion.  (I really wish Reitveld would be OK with 80 column-wide
comments!)

At least our revision history has the review links.  Those are invaluable.

-Darin

--~--~---------~--~----~------------~-------~--~----~
Chromium Developers mailing list: [email protected] 
View archives, change email options, or unsubscribe: 
    http://groups.google.com/group/chromium-dev
-~----------~----~----~----~------~----~------~--~---

Reply via email to