Re: Could timestamps be replaced with MD5?

[Tobias Weingartner]

On Tuesday, April 18, Larry Jones wrote:
> Brian Huddleston writes:
> > 
> > But if you're using any decent digest algorhytm, the statistical likelihood
> > of a digest
> > algorhytm biting you is (very conservatively)several orders of magnitude
> > less likely to bite you than timestamps.  (See my previous message).
> 
> But the consequences are several orders of magnitude more severe.  (See
> my previous messages.)

Bull.  The chances of you getting nuked out of existance by the next
big earth-quake are much higher, and quite potentially more severe...

All said, what are the chances that your CVS server gets broken in
to, and all files get posted on the web?  Or even worse, what are
the chances that all your backups get stolen, and the next day you
have a fire, wiping out all your development platforms.

I'm guessing, but I doubt I'm far off.  I'd guess that most of these
are about equal.

In some sense, you are already trusting the TCP checksums to transfer
your data correctly over the TCP/IP connection (on client server),
and the TCP checksum algorithm is a lot "weaker" than MD5, or SHA.

--Toby.