Subject: Re: Release cycles
To: None <current-users@NetBSD.ORG>
From: Greg A. Woods <email@example.com>
Date: 11/01/1997 19:58:42
[ On Fri, October 31, 1997 at 21:58:52 (+1600), Colin Wood wrote: ]
> Subject: Re: Release cycles (Was: Re: RealAudio)
> This is something of a bad idea. The nice thing about a release is that
> it usually has at least a month of stability and heavy pounding to take
> care of many of the bugs. NetBSD-current will never have this kind of
> advantage. If NetBSD is to be used in production machines, it has to be
> guaranteed to be able to run...to be stable. NetBSD-current just can't
> make this kind of guarantee.
Allow me to play the devil's advocate here for a moment:
Unfortunately no NetBSD formal release has been truely stable, even, or
especially, on the i386, since 0.9.
I commend everyone involved for the tremendous and excellent work that's
gone into NetBSD since then.
However if stability of releases is important then a whole hell of a lot
more planning and effort needs to go into the release process as opposed
to adding new and nifty features. Much of the NetBSD release effort
comes off as "#*%@! We haven't made a release for a long time! Hurry,
let's make one now!". (and if I see it this way you should see what
people who don't favour NetBSD say and think!)
Following the FreeBSD plan might be a good start.... Releases have to
have a longer lifetime (i.e. a lifetime extended with far more frequent
patch releases) and they should probably happen on a more frequent
schedule. Unfortunately this problably means branching the tree far
more often and spending a lot more time fiddling with pulling changes
from branch to branch. It's not very productive if your goal is to
advance the technology, but then neither is testing and a whole lot more
of that needs to happen in a much more rigourously planned way too.
Now of course I'm quite happy to track -current myself and be able to
drop or add patches to my own local version as I see fit. However I'd
really appreciate being able to track a truely -stable branch and then
have the freedom to pull my own features from -current as I see fit
instead of doing things the other way around.
After all these years though I get the impression that NetBSD isn't into
the O/S maintenance business in the same way FreeBSD could be seen to be
and perhaps more stable and more frequent releases aren't the best way to
spend a limited amount of primarily volunteer resources and still meet
the other project goals. If this is true though someone might try to
work on a PR campaign of sorts to try and put this in the right light
while still maintaining a high degree of excitment about NetBSD.
Greg A. Woods
+1 416 443-1734 VE3TCP robohack!woods
Planix, Inc. <firstname.lastname@example.org>; Secrets Of The Weird <email@example.com>