On Tue, 3 Aug 1999, Roy T. Fielding wrote:
> This is me learning from two completely f***ed up release plans for
> 1.2.0 and 1.3.0. It is time to do it right, and the only way we are
> going to do that is to do something different than we did before,
> with the hope that it works better. I know for a fact that this is
> the way it must be done if we are to support people doing software
> research on change analysis. I can't be entirely sure that what is
> good for that research is also good for the individual developers
> doing their own change analysis, but it is my opinion as a long-time
> developer that it is essential if we are to continue as a collaborative
> project.
I'm seeing Roy's point. Visualize it this way: if we took the current
apache-apr or apache-2.0 CVS tree, whose development looked like
1.3.x -> experiment -> revamp -> rehash -> new idea -> aha! basis for 2.0
The history for that is incredibly hard to follow. Whereas if the
history looked more like
1.3.x -> aha! basis for 2.0
where that '->' was as minimal as possible, then we've got a much better
chance of having a useful history long-term, as well as, as Roy said,
probably catching a lot of errors.
I like this; it sounds like steps 4-8 of what Roy posted would be best
done in a coordinated event, where (as someone else proposed) a group of
us would be online via IRC or something, doing this in one fell swoop.
There are steps needed to be done before this, though; and next week is
slightly tough for me since it's Linuxworld, but I'm sure between myself,
Roy, Bill, Manoj, and Dean we can find a time (and anyone else who seems
necessary).
Let me even propose a time for this hackfest: Monday, 10am PST/1pm EST?
Brian
|