httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Richards <p...@originat.demon.co.uk>
Subject Re: Voting guidelines..
Date Sun, 13 Apr 1997 14:19:40 GMT
Rob Hartill <robh@imdb.com> writes:

> that's a good thing IMO. I'd rather that someone like you or ME who
> makes few code changes can't just pop up, commit something and then
> disappear for a week or month. I think code should be committed based
> on when it's ready and not on what's a conevenient time for the submitter.

I didn't say disappear for weeks or months. It's just not always
convenient to check things in when the group dictates since few of us
work full time on this project and have to donate time as and when we
have it.

> As for voting, it has a use for anyone who wants to keep up with
> a stable version of Apache by staying in sync with the CVS tree.
> I hope more people outside of this list will start doing this in the
> future because it heads off more problems early on.
> I don't want to download a half-baked httpd one week and find
> it knocks the machine over in the middle of the night. That code
> should be robust.

This is a fruitless goal. The development tree is for development and
people should not track the cvs tree and expect a stable server. About
all they should expect is that it compiles and even that won't always
be the case.

I'd forbid people from tracking the cvs tree unless they were on this
list. You don't want people trying to run development code who aren't
following what's going on.

> Let's stop making these FreeBSD comparisions. We're doing different
> things, and their system is far from perfect too - I've downloaded my
> fair share of broken FreeBSD cvs output. I love what they do, but don't
> feel the urge to mimc their every move on the Apache front.

No system is perfect and I'm drawing on wider experience than
FreeBSD, I've worked on many software projects both commercial and
"free". I'm not driven by a desire to make the Apache development
system work like FreeBSD's, I'm driven by a desire to have a
development system that works, the current one just doesn't at all.

If you're downloading FreeBSD cvs code and it's broken then I'm really
not surprised, that's how it is if you download work in progress. If
you download a release and it's seriously broken then that would be
different (and does occasionally happen which is why there are minor
releases).

We've been drawing this beta cycle out for 6 months if not longer and
currently we have a source tree that I wouldn't trust for anything
other than a play box. I just can't see how anyone can claim the
current working practices are succeeding.

The thing that really gets me about this voting thing is that it
doesn't achieve the goals you think it does. It hasn't ensured good
code quality and has slowed down development.

If two people vote +1 on something it goes in, there's no plan as to
where the project is going and whether someone's changes fit into that
plan, there's no serious review of code changes, 2 people supporting
them is not serious code review.

-- 
  Dr Paul Richards, Originative Solutions Ltd.
  Internet: paul@originat.demon.co.uk
  Phone: 0370 462071 (UK Mobile)

Mime
View raw message