Date: Fri, 9 Jan 1998 16:12:20 -0800 (PST)
From: Dean Gaudet <dgaudet@arctic.org>
To: new-httpd@apache.org
Subject: Re: Intel Port & Release Process
In-Reply-To: <199801092353.SAA09577@devsys.jaguNET.com>
Message-ID: <Pine.LNX.3.95dg3.980109160925.21012H-100000@twinlark.arctic.org>
Organization: Transmeta Corp.
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@apache.org

It does, it's just hard to say that across multiple files unless you also
tag each commit... but tagging is not a scalable thing.  The apache-cvs
mechanism makes it really easy to back out commits, making tags somewhat
unnecessary.

A single file is trivial, Rasmus mentioned it but didn't give the
commands:

    cvs log foobar
    cvs update -r 1.second-to-last -p foobar >foobar
    cvs commit foobar

And in some cases you may want to "cvs diff -u -D 'a week ago'" to see
just what's up.  You can make a patch this way and reverse it easily.

cvs can't be told to just remove A, B, and C because B and C may overlap and
it just won't know what to do.  Humans have to help.

Dean

On Fri, 9 Jan 1998, Jim Jagielski wrote:

> Marc Slemko wrote:
> > 
> > On Fri, 9 Jan 1998, Jim Jagielski wrote:
> > 
> > > Ben Laurie wrote:
> > > > 
> > > > Marc Slemko wrote:
> > > > > The problem is that much of the code here is more interrelated and needs
> > > > > to be better on a per line basis because it simply isn't as big as some
> > > > > projects.  Voting has raised a heck of a lot of objections to all sorts of
> > > > > patches, mine included.  You can do this stuff after the fact, but the
> > > > > problem is that if no one has time to pursue it, it just falls away.
> > > > 
> > > > I'd say that the solution to this is to say that if anyone has a problem
> > > > with a particular patch then it MUST be backed out and use the
> > > > review-then-commit model.
> > > > 
> > > 
> > > >From a totally clueless person (and it might be in the FMs) how does
> > > one back out a specific commit?
> > 
> > You don't.  CVS doesn't support it.
> > 
> > You just find the diff, reverse (eg. diff the version nums in the opposite
> > directory, patch -R, etc) it and apply it. 
> > 
> 
> Oh... With all the discussion about "the power of CVS" I thought that
> somehow you could tell cvs to take out commits # A, B and C and
> cvs would handle everything else.
> 
> -- 
> ====================================================================
>       Jim Jagielski            |       jaguNET Access Services
>      jim@jaguNET.com           |       http://www.jaguNET.com/
>             "Look at me! I'm wearing a cardboard belt!"
>