On Fri, Oct 01, 1999 at 12:54:24AM -0700, longship@keyway.net wrote:
> This is what I've heard, but not what I see in the mailing list archive
> where people ask about problems with Potato and they are answered only
> "That's what unstable means." In other words, you're on your own, pal.
Well, it's a bit better than that - particularly if you keep up with the
various lists (mostly -user and -devel) you should be all right. It's
more a case of "pay attention and be prepared to fix things if they
break" than anything else. I'd guess that a fair proportion of
developers are running at least some unstable, and we like our machines
to continue to work.
> Some people just don't have the luxury of working with Unstable. However, much
> of the software released, like Gnome, GIMP, LyX and such *is* stable.
Sure, but which software and how does it play together :-) . I can
actually imagine this being a more serious problem these days with
things like GNOME having so many librarie to get right.
> So what I'd like to see is
> collection of upgrades to the current Stable from the Unstable chain, just the
> way its done in the Linux kernel. This will keep everybody happy and will delay
> the obsolescense of Stable. Right now, I wouldn't recommend Slink to anybody.
> It's just too out of date. I'm only playing around with it because I have
> a need for it in the future.
I think that's a bit extreme - this machine is running a vanilla Slink
system plus kernel 2.2 and the GNOME panel (and it's not as though I
couldn't do without the GNOME panel) and it does everything I would want
in a Unix system. There isn't much visible difference between it and
the potato systems I run.
Then again, the potato boxes are pretty much solid - my router/server
box here at home runs unstable updated every weekend, and I can't recall
any reboots other than for kernel upgrades or when it's been powered down
while I've out of town for more than a day or two. The machine which
currently acts as smarthost for tardis' outbound mail is running
unstable updated approximately daily and hasn't done anything
particularly nasty to me.
If you really want to run an up-to-date system and can't tolerate any
breakage at all then you probably want to have a test box sitting by
which you can try out the new versions on before your production system
falls over.
Otherwise, giving it a day or two before installing new packages and
paying attention to bug reports and the lists should help you steer
clear of anything really serious. Using apt, you can track just the
list of packages you need rather than the entire distribution.
> What can we do to get this accomplished?
> I'm willing to put in some work to get this ball going. Does anybody
> else see this as worthwhile?
Check out the extensive existing discussion in the -devel archives. There
are several proposals, some of which would require code to be written.
IMHO it's probably enough to get the release cycle down to six months,
though a semi-stable distribution may be one way to achieve that.
--
Mark Brown mailto:broonie@tardis.ed.ac.uk (Trying to avoid grumpiness)
http://www.tardis.ed.ac.uk/~broonie/
EUFS http://www.eusa.ed.ac.uk/societies/filmsoc/
Attachment:
pgpvoaJarn2_e.pgp
Description: PGP signature