Re: error message from CVS [CLOSED]
On Tue, Jul 16, 2002 at 04:07:44PM +0100, Shri Shrikumar wrote:
> On Sun, 2002-07-14 at 22:38, Paul E Condon wrote:
> > I'm attempting to learn CVS to use in keeping track of my own work.
> > I had it working about a month ago. I went off to do other work, and just
> > got back to using it today. Now I get the following:
> >
> > pecondon@big:~/co/s2$ cvs commit
> > cvs commit: Examining .
> > cvs: hash.c:312: findnode: Assertion `key != ((void *)0)' failed.
> > cvs [commit aborted]: received abort signal
>
> I can only recommend upgrading CVS, checking you CVSROOT for any
> corruptions etc. It might be a bug I suppose. Also, it might be better
> to try one of the CVS mailinglists which can be found at
> http://www.cvshome.org
>
>
> HTH,
>
>
> Shri
>
Since posting my last request for help, I have recalled some important facts
about this problem.
I had attempted to commit the s2 project earlier in the day and had gotten
an error because I ran out of disk space on the partition where my repository
is located. I deleted some stuff on that partition, got distracted by other
work, and came back to doing a commit later, and got the error message that
I posted.
On remembering this, I've done some more research.
I tried checkouts and commits of other projects in the repository.
They all work, error free.
I removed the s2 directory from the repository, and saved it elsewhere on
disk for further investigation. I then created a new s2 project by importing
from my checkout area.
It works, but lacks the earlier history of the s2 project.
I believe that cvs has bugs in two places: Its error recovery for
out-of-space on disk, somehow deposits garbage data in the project records
in the repository. (bug 1) and (bug 2) cvs commit sets a null value in a
variablae, 'key', because of the garbage and then bails out on an assert
rather than issue an error message that is informative to a dumb user.
(Maybe this really isn't a bug. Should a dumb user be attempting to use CVS?)
I attempted to save copies of all the files involved the the old, error-
producing situation. I hope to be able to reproduce the assert error. If I do
reproduce it, I'll submit a bug report.
Thanks for puzzling over this.
I'm sorry for not having given you all the relavant information.
--
Paul E Condon
pecondon@quiknet.com
--
To UNSUBSCRIBE, email to debian-user-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: