--=-=-=
Content-Transfer-Encoding: quoted-printable


Thanks for all the comments.

I should have said: I'm talking about my notebook, which is running
current from April.  I'm aware of netbsd-3, and run that (or netbsd-2
still) on a number of production machines.  My project at BBN will be
tracking current, so it's ok if this takes a bit of time but I'd
prefer to avoid real trouble.

So my question was really: is -current in the mostly ok shape it
usually is, or in one of those occasional bad states?  From the
responses I got, it seems yes, it's in reasonable shape.

  From: Daniel Carosone <dan@geek.com.au>

  Seems to be to me, and there are many useful fixes. Certainly, no
  worse than -current ever is, and a lot better than it sometimes
  (rarely) can be.

Thanks, and that's what Garrett said mostly.

  > I've been
  > hesitant to upgrade due to the hearing about the switch to gcc4.

  This begs two questions:

   - upgrade from what as a starting point?  an earlier current from
     (say) a few months ago, or from a 3.0-something?  if the former,
     i'd say go ahead and swap a whole bunch of known fixed issues in
     the past few months for the risk of new ones that you will always
     get with -current, and faced the last time too.

From=20April current, so seems like I should go ahead.

   - what about the switch to gcc4 makes you hesitate?  The fact that
     gcc4 might introduce or uncover additional problems (in which case,
     don't worry, because it's not yet the default), or the fact that
     you may want to upgrade again very soon once the default is
     changed, and you hesitate so as to wait and do the work only once?

I use a script that does full-auto merging of etc, so the work is only
dealing with any problems
(https://acert.ir.bbn.com/projects/etcmanage/).  With gcc4 not the
default, I'm not worried.

  As an aside, my concern with switching to gcc4 relates more to the
  possibility of random pkgsrc things breaking than to the base system;
  accordingly, I have so far only switched machines that don't use many
  pkgs to gcc4 - and have had no problems with those.  When I have some
  spare time I'll knock up another xen domU to test building all the
  pkgs I use elsewhere with gcc4, and see how many break.

That's a good point - my thinking was more diffuse, that upgrading the
compiler was scary and that I didn't want this box on the profusely
bleeding edge.

One person said offlist that a build with HAVE_GCC4 completed and the
kernel crashed on boot.  I'm used to having to cvs up the next day and
retry the build - my real concern was if a new build would succeed to
make release but then work very badly.

=2D-=20
        Greg Troxel <gdt@ir.bbn.com>

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)

iD8DBQFEkBcp+vesoDJhHiURAoqnAJ9vrqVbl++Y6iCicpVKxieWBivF7wCaA6Hq
I3N8LI51yTWNDuUPkFvoJKQ=
=Fj9e
-----END PGP SIGNATURE-----
--=-=-=--