> I find it much safer and easier to use tags and the patch(1) command.
> I've posted something similar to this procedure in the past to the
> info-cvs mailing list, but I don't think it's in the FAQ or
> documentation (yet):
> 

Well after a few days no one has jumped all over Greg (or myself) so
obviously no one feels very strongly either way? 

> I find it much safer and easier to use tags and the patch(1) command.

I like patch too :-) Does your procedure involve more munging over the
CVS tree or less?  A sparc is not toooo bad, but my poor i486DX33 at
home takes _hours_ to do a full import/merge etc.

> 	# this step is a bit heavy, and only necessary if you want to be
> 	# really pedantic and only move the default branch back on true
> 	# current files (i.e. not on locally modified but out-dated files)
> 	cd /var/tmp
> 	cvs co -r NetBSD-950603 NetBSD
> 	cd NetBSD
> 	cvs admin -bNetBSD . >/dev/null	# WARNING: no space after the '-b'

Not 100% sure I followed that...

> 	cvs admin -bNetBSD . >/dev/null	# WARNING: no space after the '-b'

Makes NetBSD the default banch... in the checked out sources.

> 	# now update the working tree to 950603 -current
> 	cd /usr/src
> 	cvs update -A -ko . | tee ../update-950603 2>&1

Now you forget the -bNetBSD? or does this only affect new files?
If not, what was the point of the -bNetBSD ?

The -ko seems a good idea.  Though if I edit a file it should have 1.2
not whatever the old id said... also since NetBSD use $NetBSD$ rather
than $Id$, my CVS does not touch them...

Also, if I edit a file and commit it does it go on the vendor branch
or on the head?

> 	cvs -nq tag -F NetBSD-950603 . | awk '{print $2}' > ../removed-950603

Sounds like I need to update my CVS, my docs don't mention '-F' what
does it do?

> NOTE:  I've not yet done this on the NetBSD tree, though I've done it
> several times on CVS, BIND, Smail-3, and a few smaller things.

As I indicated I've done the cvs merge stuff on NetBSD a few times
with no problems.   But it takes a _long_ time.

--sjg