Re: code reviews

[Ben Pfaff]

John Darrington <address@hidden> writes:

> On Sat, Jul 01, 2006 at 01:23:41PM -0700, Ben Pfaff wrote:
>      Let me elaborate.  I've watched three different systems of code
>      reviews in some detail:

[...]

> Have you looked at the way aegis works?  One of its features is that
> it can require code reviews before anything gets checked in.  The
> review policy is configureable according to the enterprise's needs.

I don't want to enforce code reviews.  I want them to be
optional, but useful enough that contributors choose to get them
done voluntarily.  If they are not useful, then we shouldn't
waste our time on them.

>      > The "code reviews" conducted at any commercial organisations with
>      > which I've been involved, have been a joke.   Nobody had the
>      > time/competance/inclination to do them.  Consequently, if there were
>      > done at all, they were simply a beaurocratic exercise.
>      
>      It sounds like your own experiences with code reviews have been
>      the opposite of mine.  I'm sorry to hear that; they can be really
>      good things.
>
> I didn't mean to sound quite so negative.  I agree that code reviews
> can help.  But I've seen so many software houses, where (usually to
> justify a claim of compliance to ISO-9001 and/or CFR 21.11) the
> directive is given "henceforth all software changes must be reviewed
> (and recorded in form XYZ)".   It didn't improve software quality and
> just became an administrative burden. 

Ugh.  I would not want to work in such an environment.

> I like the idea that author of the patch gets to make the decision to
> review/not to review, and to accept or decline suggestions from the
> reviews.  I think it's also a good idea to suggest that patches which
> have not been reviewed after say 7 days can be automatically committed
> at the author's discretion.

Sure.
-- 
"GNU does not eliminate all the world's problems,
 only some of them."
--Richard Stallman