> > The CVS server seems to be working again, but logging in with an
> > empty password doesn't work. The web interface to anonymous CVS
> > doesn't work either.
>
> > [bruno@bruno bruno]$ cvs -d :pserver:anoncvs@anoncvs.postgresql.org:/projects/cvsroot login
> > (Logging in to anoncvs@anoncvs.postgresql.org)
> > CVS password:
> > /projects/cvsroot: no such repository
> > cvs [login aborted]: authorization failed: server anoncvs.postgresql.org rejected access
>
> I believe the intention is to get rid of the separate anoncvs mirror
> and make anoncvs access go directly to the primary archive. This
> will have two good effects: eliminate the synchronization delay
> between primary and anoncvs repositories, and eliminate the (bogus)
> difference in cvsroot paths, which creates unnecessary diffs between
> what committers and noncommitters see in their checkouts. The bad
> effect is that noncommitters will have to re-check-out from the
> correct repository path, which is just /cvsroot.
Actually, they won't have to check things out if they run:
find . -type f -name Root -path '*CVS*' -exec perl -p -i -e
's#:pserver:anoncvs\@anoncvs.postgresql.org:/projects/cvsroot#:pserver:anoncvs\@anoncvs.postgresql.org:/cvsroot#go'{}
\;-print
Moving to consolidating the anoncvs server as the same as the master
repo will have the bad effect of being quite hard on the disks when
people do cvs update's (if anyone cares). Remote updates are quite
inefficient/hard on the cvs server (reason FreeBSD requires committers
to CVSup and use their local copy instead of cvs up'ing off of the
master).
-sc
--
Sean Chittenden