> tree, I'm proposing that it would be a good idea to bring CVS into
> the main tree, rather than having it just a package. If people are

Unless there are licensing problems, I'd be in favor of that.
Who knows, it might improve cvs a little. (/tmp file philosophy?
Are we supposed to have one of those?)

> worried about `bloat,' I'm happy to consider proposals for removing
> other, less-used tools, such as RCS, or perhaps sup.

Um, er, I think CVS uses RCS internally... Certainly it didn't work on HP/UX
until I downloaded GNU RCS, installed that, and rebuilt CVS on the HP...

As for sup, it's FAST when you update often and only care about tracking the
head of the main branch, because that's what it is optimized to do. CVS will
grind through the whole tree which is murder for slow links; isn't this why
CVSup exists?

Bottom line: I will SUP NetBSD through my ISDN line, but I won't run anoncvs
checkouts or updates through it. I take the sparcbook into work when I need
to rsync the public repository or do an anoncvs checkout. (Yeah, yeah, DSL
someday, argh.)

> are the two things I *always* find myself building whenever I
> install a NetBSD system for any purpose.

These days my list is tcsh, cvs, rsync, ssh...

Another argument in favor: people contributing to open source in general
will probably be using cvs, so we most likely have a good number of users
who are already installing it themselves because it is not in the tree.

Now the flame war: base or comp?

Todd Whitesel
toddpw @ best.com