On Sun, 20 Apr 2003, Havard Eidnes wrote:

> I'll have to side with those who oppose your proposed new numbering
> scheme.
>
> First off, it seems that people who have a difficulty with trying to
> compare 1.6.1 to 1.6L or 1.6O (or whatever version of NetBSD-current
> produced after the netbsd-1-6 branch was created), or who try to
> "upgrade" from 1.6R to 1.6.1 (which will of course fail miserably) have
> not really grasped what creating a "branch" means, and that it's
> possible (and customary) to do maintenance on such a branch by applying
> selective fixes on the branch independent of the development which is
> happening in -current, and that such maintenance continues for a Long
> Time after the initial major release.

Why do our users need to understand what a "branch" is? Sure, the
difference between "development" and "safe" is important, but why do they
need to understand things the way the developers do?

> It's also clear that those who try to "upgrade" from 1.6R to 1.6.1 have
> not read the first part of the install notes either, in particular the
> section "Upgrade path to NetBSD 1.6.1".

True. But why should we punish them for it?

It's a recurring problem, why not come up with something that makes sense
both to developers and to new users? Or at least closes the windows for
foot-shooting?

> I get extremely queasy with the introduction of "magical numbers with
> special meaning" in your proposed version numbering scheme, and frankly,
> don't really see what it helps against.  Instead, the rule "contains
> letters, is not a release" should be quite straight-forward to
> understand.

It should (and is for many), but it isn't for a lot of folks.

While I'm not saying this method is what we should do, I think we should
do something different than what we do now.

Take care,

Bill