Subject: Re: mono experience cutting to subversion
To: Pavel Cahyna <pavel.cahyna@st.mff.cuni.cz>
From: Greg A. Woods <woods@weird.com>
List: current-users
Date: 02/12/2005 15:49:37
[ On Friday, February 11, 2005 at 23:57:21 (+0100), Pavel Cahyna wrote: ]
> Subject: Re: mono experience cutting to subversion
>
> The feature you mention isn't supported by CVS - it's implemented by CVSup
> "above" CVS. How is SVN worse than that?

svk isn't really anything like CVSup -- especially not yet, and it's
also not (yet) oriented towards mass distribution.

Of course svk _is_ something along the lines of trying to mangle
subversion into something more like BitKeeper, DARCS, et al, but it
seems to do so at the wrong "level".  If you/we want to move to a
changeset based development model then why not go directly with a mature
versioning system that already works that way at its very core?

I think svk would also need a full rewrite before it would really be
usable on a larger scale.

I agree fully that CVS has limitations, some serious, but its
limitations are _NOT_ what's causing problems with distributing the
NetBSD source repository.  On the contrary subversion would seem to
suffer even larger problems with mass distribution of its repository.

Further the CVS limitations are already well known to NetBSD developers
and there are "adequate" procedures for working within them.  FreeBSD's
repository has maybe an order of magnitude more tags, branches, etc.,
than NetBSD's, as well as quite a few more files IIRC, and although
they've made more rumbles about ditching CVS, they still haven't done
so.  See also the thread beginning:

	http://lists.freebsd.org/pipermail/freebsd-hackers/2004-February/005592.html

Please let's try _not_ to fix what is not broken.

-- 
						Greg A. Woods

H:+1 416 218-0098  W:+1 416 489-5852 x122  VE3TCP  RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>