On Wednesday 27 September 2006 17:39, Peter Samuelson wrote:
> [Ph. Marek]
> > What about the use-case
> > cp -a foo1 foo2
> > (edit foo1)
> > (find that changes are bad)
> > mv foo2 foo1
> > That wouldn't change the data, but the inode.
>
> That's fine. Requiring a full-text compare where it is not needed is a
> lot better than _not_ doing a full-text compare where it _is_ needed.
> This is why we have 'svn cleanup' which scans the WC and resets those
> parameters, so you get your optimisation back.
IIRC you have no guarantee that the inodes are constant when using a WC on
NFS. So you'd change the behaviour
(same mtime, same size) => no checksumming (like now)
to
(same mtime, same size, different inode#) => checksumming,
which, IIRC, may be unacceptable (speed-wise) for NFS.
(GNU arch :-) has such a description:
http://www.gnuarch.org/gnuarchwiki/NfsWorkplaces)
Regards,
Phil
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Sep 28 07:17:08 2006