httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Richards <p.richa...@elsevier.co.uk>
Subject Re: is 1 week enough?
Date Wed, 14 Feb 1996 11:31:10 GMT
>
> I'd like to ask that the cvs-admin (and anyone else involved in setting
> up the new process) *please* keep a living document, in our developer
> space, which explains the current state of whatever we believe the
> process to be.  It cannot be improved if we don't know what it is.

I'm in full agreement with this but I don't want to unilaterally write
policy documents. I want this list to determine what that policy will
be. It probably won't be everything I want or everything Rob wants or
please everyone but I hope it will move in the right direction and I'll
certainly pipe up with aspects of policy that are governed by cvs use.

> One thing that we should note is that the current patch system forces
> at least three people (and usually 6-7) to look at each patch.  This
> overcomes our natural habit of only looking at something when it breaks,
> and thus catches more non-compilation problems earlier in the cycle than
> I have seen with other projects.  Even with voluntary peer review,
> we won't get an equivalent forcing-function out of CVS -- we may have
> to come up with a new check-point just to compensate.

CVS doesn't impact peer review at all really. It's voluntary now since I
never review patches, I just grab the latest patched updates and if I find
something broken I'll complain then. With cvs I'll probably update my local
copy every time a commit is made (or certainly every day) and compile it
there and then so I'm always running the latest and greatest version so
at least in my case I'll be doing more peer reviewing than I do now.

cvs makes this less work to do since I don't have to go and ftp patches
and apply them all the time, dealing with the conflicts and so forth. I
just have an up to date version arrive on my box automatically which I
can just compile and start using straight away. I can also go in and
*fix* problems straight away instead of maintaining a locally hacked version
that works for me and then have to deal with the hassles of providing a 
patch against a clean version of Apache of my changes while still coping
with other people's patches that clash with my changes.

As I said, cvs is not a project management tool, it's maintenance tool but
lokking after the source is an important aspect of project management which
is why tools like cvs exist in the first place.

Mime
View raw message