Re: Checksum failure: serious problem or not?

[Aad Rijnberg]

On Tue, 2003-12-23 at 21:14, Eric Siegerman wrote: 
> Or the client file changed but its timestamp did not.

Hi,

Thank you all for your explanation so far.

I was the one who started this discussion. I am just a normal CVS-user
(also playing CVS-administrator for our project team), so I have maybe
read (a little) more than the average CVS-user.

When I got this message I thought that there might be some problems with
the repository, which might eventually lead to a corrupt "database" (in
Visual Source Safe this can happen once in a while...). That is of
course something I would like to prevent. So that's why I wanted to know
how serious this warning was.

For a long time we have used CVS without any of those warnings; it
popped up only recently. After the remark of Eric I think I know why we
have these messages now. One of my colleagues wrote a script to remove
tabs, CR/LF, and beautify the code with indent. In doing so he chose to
leave the date/time stamp of the file equal to that of before the
beautifying action. His reason was that if people did not modify a file,
then it is not desired to check in a file just because it was
reformatted. After following the discussion, I can understand that the
patch will not work on a file which has been changed locally, which CVS
is unaware of (same time stamp). Then the file is completely downloaded
from the repo, and overwrites the locally modified file.

Am I right that this is probably the cause of the problem?
If you have any suggestions for improvement in our approach, please feel
free...

Kind regards,

Aad Rijnberg