Re: [HACKERS] Current sources?

On Mon, 25 May 1998, David Gould wrote:

> Where I work we have had good success with the following:
>
>  - every night a from scratch build and regression test is run.
>
>  - if the build and regression is good, then a snapshot is made into a
>    "last known good" location. This lets someone find a good "recent" source
>    tree even if there is a problem that doesn't get solved for a few days.

    Actually, ummm...I've been considering removing the snapshot's
altogether, now that anoncvs works.  The only reason I was doing it before
was that not everyone had access to CVSup for their particular
platform...the snapshots are a terrible waste of resources, especially
considering that you have to download all ~3.5Meg (and growing) .tar.gz
file each time...whereas anoncvs/CVSup only updates those files requiring
it...

    IMHO, the snapshot's are only important during the beta freeze
period...

> The other tool I believe to be very effective in improving code quality
> is code review. My experience is that review is both more effective and
> cheaper than testing in finding problems. To that end, I suggest we
> create a group of volunteer reviewers, possibly with their own mailing
> list.

    That's kinda what pgsql-patches is/was meant for...some ppl (I
won't mention a name though) seems to get nervous if a patch isn't applied
within a short period of time after being posted, but if we were to adopt
a policy of leaving a patch unapplied for X days after posting, so that
everyone gets a chance to see/comment on it...

> I don't have strong preferences for the form of this, so ideas are welcome.
> My initial concept supposes a group of maybe 5 to 10 people with some
> experience in the code who would agree to review any patches within say two
> days of submission and respond directly to the submitter. Perhaps somehow the
> mailing list could be contrived to randomly pick (or allow reviewers to pick)
> so that say two reviewers had a look at each patch. Also, I think it is
> important to identify any reviewers in the changelog or checkin comments so
> that if there is a problem and the author is unavailable, there are at least
> the reviewers with knowledge of what the patch was about.

    This sounds reasonable to me...this is something that FreeBSD does
right now...one of these days, I have to sit back and decode how they have
their CVS setup.  They have some things, as far as logs are concerned,
that are slightly cleaner then I have currently setup :)

> I would be even happier to know that next time I had a tricky patch that
> some other heads would take the time to help me see what I had overlooked.

    You always have that...:)