(fwd from uke@jeremy.org) thanks and patch

[jeremy bornstein]

On Thu, Mar 21, 2002 at 10:07:14AM +1100, Martin Pool wrote:
> It sounds like you're using asymmetric encryption.  So I suppose every
> time you encrypt the file, gpg will generate a new session key, so an
> identical cleartext file will generate a completely different
> cyphertext file every time.

Yes, this is correct.


> On 20 Mar 2002, jeremy bornstein <uke at jeremy.org> wrote:
> > > Wouldn't encrypting the file with gpg change the timestamp as well as the
> > > size, so rsync would still copy the file?
> > 
> > It certainly does--which is why I reset it afterwards.
> 
> Why not just re-encrypt the file only if it has changed since the last
> transfer?  You could do that either by keeping the encrypted file on
> the origin machine, or by using rsync to look at the modification time
> on the remote machine.

Yes, this is what I do: use rsync to look at the mod time on the
remote machine as compared to the mod time of the original
(unencrypted) file on the local machine.  I can't keep the encrypted
files because of disk space limits.  Files are encrypted only if rsync
tells me that the local file has changed.  (This is what the patch is
for.)

> (I'm not just saying this to be difficult.  We can't merge patches
> unless there's some reason to believe people would actually use them,
> or otherwise the code will become a complete mess.)

Of course not--I completely understand.  I suspect that my use is not
unique, but I don't have any supporting specifics and I won't discount
the possibility that it's not worth supporting in your software.

-j