httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Laurie <...@gonzo.ben.algroup.co.uk>
Subject Re: Voting Summary
Date Wed, 06 Dec 1995 11:12:19 GMT
> 
> In reply to Randy Terbush who said
> > 
> > > On Tue, 5 Dec 1995, Paul Richards wrote:
> > > 
> > > > On the 1.1b1 stugg, why not just do what all the free BSD projects do
and
> > > > have an Apache-current, which is never frozen and is constantly at the
> > > > cutting edge of development which everyon sticks their new ideas into.
> > > 
> > > Correct me if I'm wrong, but wouldn't this extremely mess up our system 
> > > of patch-and-vote? It wouldn't give the rest of the group any say as to 
> > > what any one person is doing to the code.
> > 
> > I agree with Alexei's comments here. I've been tracking a -current OS
> > for about 2 years now. It is a mess (if not impossible) to track back
> > to when an errant change was made, etc. For a source tree as big as
> > an OS, a patch system is difficult to pull off. Luckily, our source
> > tree is not that big... yet.
> 
> Not if you have access to the CVS tree :-)
> 
> Hmm, I think the patch and vote thing stinks anyway which is why I'm not
> contributing any code. The changes I want to make (cleaning up all the
> #ifdef's for a start) will touch every file, include some new ones and
> remove some old ones. It's *extremely* unlikely I'd get it right for
> every OS in my first attempt and the patch and vote system falls
> flat on its face trying to do things like this.
> 
> It took us something like 2-3 weeks to get 3 or 4 lines changed for
> just one #ifdef last time because the patch didn't work for everyone and
> it was voted down.
> 
> A patch system works when the changes are just that, patches.

I think the patch and vote system works well for "final" releases, where it
is vital that everything works everywhere.

I find it difficult to imagine how your continuously changing source works. If
I, for instance, go and sort out this connection abstraction thing, which
touches almost every file, and at the same time you sort out the #ifdefs, how
do we reconcile the two sets of diffs?

I can see it for an OS, coz you can say "hands off this subsystem for a couple
of weeks while I sort out X", but you can't say "hands off Apache for a couple
of weeks". Or can you?

It has occurred to me that for sweeping changes, there should be a special
1-patch version, which belongs to the maker of the sweeping changes. Obviously,
whoever does this should try to do it quickly!

I'm in favour of relaxing the voting rules for the beta releases - for
instance, we could allow OS incompatibilities (so long as someone fixes them
in the next round).

Cheers,

Ben.

-- 
Ben Laurie                  Phone: +44 (181) 994 6435
Freelance Consultant        Fax:   +44 (181) 994 6472
and Technical Director      Email: ben@algroup.co.uk
A.L. Digital Ltd,           URL: http://www.algroup.co.uk
London, England.

Mime
View raw message