On Thu, Dec 04, 2003 at 12:17:49PM -0800, Jason Thorpe wrote:
> 
> Not without screwing over basically the entire user-base.
> 
Can one not fork a new, incompatible branch and provide a migration
path from the legacy wherever/whenever possible?  In other words,
provide two distributions and deprecate the legacy one while
providing tools/documentation that facilitate migrating from the
deprecated version to the new one when legacy applications become
obsolete.

I appreciate the need to support the user base indefinitely, but
eventually one expects the legacy stuff to dwindle into effective
insignificance unless we encourage, as we seem to do now, hanging
onto obsolete software.

Also, I understand this is not by any means simple, but I believe
it needs open minded discussion and a commitment to move on before
NetBSD paints itself into a corner.  Talking about understanding,
I'm sure I _don't_ understand all the ramifications of changing
libc's version, but I can't see how we can burden all future use
of it, may it live a long, long time, with a legacy already of
quite a few aeons, in computer terms.

And, no, I am not soliciting flames and like many others here I
appreciate how much effort is going to be required to change path.
But I cannot believe that the pressure to move off that path is
not going to become greater than the NetBSD project can withstand.
When that irresistible force hits the unmovable NetBSD, I will be
extremely concerned to be one of the casualties.  I and my clients,
the latter being an even more serious concern.

++L