[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: "server-less" local modification detection?
From: |
Larry Jones |
Subject: |
Re: "server-less" local modification detection? |
Date: |
Thu, 13 Jun 2002 11:33:32 -0400 (EDT) |
Thomas Nussbaumer writes:
>
> Suppose you don't have always a connection to your CVS server, but you want
> the client to check if a file is locally modified. Time and size based
> checks are not sufficient for this task. What you need is some kind of
> checksum (MD5) which you can check against modifications. In the best case
> these checksums get stored in file CVS\Entries AND may be embedded as tag in
> sourcefiles (would enable runtime checks within the software).
>
> Would it be hard to integrate this feature into the CVS sources? I'm not
> requesting an official feature, but I just want to discuss the technical pro
> and contras ...
Adding a checksum to the Entries file shouldn't be too hard and has been
suggested a number of times as much more robust than just using
timestamps as CVS currently does. Upward (and downward?) compatibility
could be a bit tricky, though, and there's always the potential for a
modified file to end up with the same checksum (although with a large
checksum like MD5 the probability of that happening is essentially
zero).
The more complicated part would be the purely-local status check -- the
standard CVS client always sends any potentially-modified file to the
server to find out for sure, there aren't any provisions for running
without the server available.
-Larry Jones
Moms and reason are like oil and water. -- Calvin