Subject: Re: detecting -current versions
To: Brian R. Gaeke <brg@dgate.org>
From: David Maxwell <david@spinne.web.net>
List: current-users
Date: 09/05/1995 15:19:25
Quoting  Brian Gaeke:  Internet E-Mail: <brg@dgate.org>
> And then spake Postmaster@spinne.web.net as follows:
> > The name is rather important. For one particular reason. Interim, or 
> > Pre-release sounds like an 'official' release of sorts. To me anyway,
> > virtual doesn't have the same connotation. [...]
> How about 'Unofficial', then, with version numbers like 1.0u1?

That's okay with me, but it's still releated to 'official' in my mind. I'm
thinking of the difference between moral, immoral, and amoral. I want an
aofficial release ;-)

> Well, people *do* already put up binary snapshots on their own. From what
> I've gathered there's not all that much difference between a reasonably
> stable binary snapshot and an unofficial release; on the mac68k side we
> have the "wormey tarballs" and the "puma snapshots", which lots of people
> run because so much progress has been made since 1.0. Am I missing the
> point here?

I'd like to see the wormey and puma things come under one heading, preferably
cross platform too. I know we won't be able to generate completely 
in-sync releases for every platform, but the numbering scheme should not
give the same number for different platforms build from different sources.
idento avoid what Brian Gillham mentioned (in another message) about John,
Dick, and Harry's Linux releases.

> > could be more self-supporting. With a well designed web page, people could
> > look up known problems based on their current V-rel. This could even tie in
> > to PR's, which could be fixed as of a certain V-rel.
 
> This would be good, but keeping track of the fixed PRs in each unofficial 
> release and generating release-to-release patches could get hairy.

It'll be work, but as I said in a previous message, this is one way some less
technical people could contribute to NetBSD. (Myself included ;-)

> All good points. And people that don't want to run -current wouldn't be
> stuck running something ancient.
> 
> > The way snapshots are made, and who makes them now would not change, we
> > would just add more structure to the way they are made available. I can
> > probably provide a Web site for doing the things mentioned above.
> 
> What about deciding when to make them? (graph the send-pr's/week, and
> release them at reasonable minimums. :)

I'd say make a snapshot, but don't release it the same day. Watch the mailing
lists for a week (or some set time), and release it as is, or with submitted
patches that have shown up.

							David Maxwell
							david@web.net