Hi Rob!

In message <9611061526.ZM11796@warhol>
          Robert Black <r.black@ic.ac.uk> wrote:

> > The last line doesn't make sense at all IMHO because you can't base
> > something release-like on a moving target like NetBSD-current.
> 
> The idea is that if you find a kernel which is particularly stable you make a
> release set based on it, much like FreeBSD.

In this case, it would IMHO be the best solution to use another minor number,
ie. the next time new base sets etc are built from NetBSD-current sources, I
would call them RiscBSD 1.2.1.

> The only bit of X which changes rapidly is the server. THe major changes tend
> to be to support new versions of the operating system (changes in kernel
> drivers, etc). This means that to avoid confusion the X release needs to be
> tied to the OS release in some way. I really ought to use an OS-based number
> for the versions of Xarm.

You're right, this makes a lot of sense.
But how do you name the kernels in a sensible manner? Most people are probably
using pre-compiled kernels, but they're now split into StrongARM, voyager and
a7000 kernels.
Others such as I prefer to compile own kernels from suped sources.
Now it is not easy to work out which compile number as used by Mark a given
suped source tree corresponds to, especially as not all experimental features
are checked in to the tree immediately.
If I were you, I would still use Mark's number though. Sup users normally
notice what changes have been made and will thus know the corresponding pre-
compiled kernel. Although this is not ideal because it's rather difficult, I
don't see a better solution atm.

MB