Re: cvs update: checksum failure after patch

[Larry Jones]

Peter Toft writes:
> 
> I have a nasty problem now with a CVS repository, where I can update after 
> some other user has committed - now I get this:
> cvs update: checksum failure after patch to <myfile>; will refetch
> 
> Is it bad? How bad? How to rework around it?

It's a little bit worrisome, but not too bad.  Normally, what that means
is that your local file had been changed without changing the timestamp,
so CVS didn't know that it had been changed.  Since CVS didn't know the
file had changed, the server just sent the differences between the
version you had and the new version and the checksum for the new
version.  After applying the changes, CVS computes the checksum for the
local file and compares it against the checksum the server sent to
verify that the local file matches the server file.  In this case, the
checksums didn't match, so the local CVS went ahead and fetched the
entire file from the server, so your local file now exactly matches the
server's file.  Of course, that means that the undetected changes have
been lost.  If there weren't supposed to be any local changes, there's
nothing else you need to do.  If you really did want those changes, the
old file should have been saved with the name changed to start with ".#"
and end with the version number (e.g., foo.c becomes .#foo.c.1.23).

The important thing is to figure out how the file got changed without
changing the timestamp and make sure it doesn't happen again, since you
can easily lose changes that way.  If it happens all the time instead of
just being a one-time thing, there's something seriously wrong with your
CVS setup.

-Larry Jones

I wonder if you can refuse to inherit the world. -- Calvin