Subject: Re: NetBSD 2.0 release date
To: Jason Thorpe <thorpej@wasabisystems.com>
From: Greg A. Woods <woods@weird.com>
List: tech-kern
Date: 12/07/2003 13:21:45
[ On Saturday, December 6, 2003 at 11:15:27 (-0800), Jason Thorpe wrote: ]
> Subject: Re: NetBSD 2.0 release date
>
> Yes, it is.  If you bump libc, you have to bump the major of every 
> library that exists that uses libc functions, including 3rd party 
> libraries.

Agreed.  Such are the limitations of shared libraries and in particular
the versioning schemes used for them by NetBSD.

> This is simply too large of a task.

I disagree with this very _strongly_.

Neither you nor anyone else have given any even half-educated guesses to
the questions I've asked in order to try to quantify this issue.

Your gut feeling is that it will cause "great great pain" -- too much
pain to make the effort worthwhile.

My gut feeling is that it won't cause any significant pain to any
significant number of people.

No doubt the truth is somewhere in between but regardless the proof is
already available from at least one major commercial unix vendor (and no
doubt more too) that even with many orders of magnitude more proprietary
third-party code as NetBSD has such upgrades can be done without
incurring so much pain as to kill the patient.  While those commercial
vendors have been able to dedicate significant resources to the task I
believe they have had to do so only because they directly and
contractually support all that additional proprietary third-party code.

-- 
						Greg A. Woods

+1 416 218-0098                  VE3TCP            RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>