On Mon, 2004-11-01 at 16:19, Ron Johnson wrote:
> On Mon, 2004-11-01 at 15:50 -0800, Dan Stromberg wrote:
> > On Mon, 2004-11-01 at 15:14, Jon Biddell wrote:
> > 
> > > >
> > > > If I recall correctly, this was considered and not implemented-- it's
> > > > not clear that nearly-identical messages can be identified properly
> > > > without a lot of processing.
> > > 
> > > Interesting - I wonder how the kmail guys do it - it seems to work pretty 
> > > well, with only the rarest of false deletions - I had a mailbox with 16k 
> > > messages (deliberately created to test it) which was a double-import of 
> > > another mailbox - I *knew* there would be exactly 8192 duplicates, and kmail 
> > > shot through the file in less than 30 seconds.
> > 
> > If you want to be quick, you can just delete all but one copy of
> > anything with the same Message-id: header.
> > 
> > If you want to be more thorough, you could additionally generate sha-1
> > or md5 hashes of all messages as they come in, perhaps inserting them
> > into a heap.
> 
> That's the exact idea I had, except I was thinking of bash and
> Maildir...

Yeah, that's good too.

BTW, deduping the addressbook would be nice as well.  A lot of my
addressbook entries have the same e-mail addresses twice - probably as a
result of a failed experiment in which I tried to migrate from jpilot to
evolution for syncing with my palm.

-- 

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to