On 27/08/06, Bob Proulx <bob@proulx.com> wrote:
> mal content wrote:
> > No, it doesn't exist, and I really have no idea what could have removed
> > it. I don't use any tmp cleaners or anything like that. My exact actions
> > were to run make, run make clean, edit two files, run make and then
> > attempt to commit. Nothing in any of those invocations of make would
> > have touched that directory, so this is a little worrying.
>
> Is the working copy over NFS? I ask because I have seen similar
> problems with the .svn directory becoming corrupted but only when the
> working copy exists on an NFS filesystem. It always works fine to a
> local filesystem. I have been thinking this was a linux kernel nfs
> bug and have investigating but so far no root cause of it.
>
Heh, no, I have enough trouble without throwing NFS into the mix as well!
> > For now, I'm just going to create a new branch from the last commit
> > and copy diffs over.
>
> It is also possible to do a clean checkout elsewhere and then restore
> the .svn directories from the clean area. That would replace your
> corrupted .svn dirs with known good .svn dirs. I have done this when
> I have had this NFS-related problem. That operation is a little
> tricky and you must be careful especially if there is a lot of
> hierarchy of directories involved. Don't get them mixed up. But if
> you have a lot of changes outstanding may be easier than other
> alternatives.
Sounds terrifying. Don't suppose there's a script somewhere that can
do this? I'd rather not bank on getting it right first time.
MC
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Aug 28 05:43:50 2006