More robust INDEXing

[Ken Murchison]

Forrest Aldrich wrote:
> Slightly OT: however, there are a few programs out there that offer a 
> richer indexing method (a la Google for your email).
> 
> I wonder if there's a place for that somewhere in the Cyrus structure. 
> There's obvious utility in a richer search-and-find method.
> 
> I'm probably not alone with my hoard of emails, spanning many years (and 
> many memories), amounting to thousands (or more!) of individual messages.
> 
> I have these all slightly organized in my Cyrus IMAP spool, which I back 
> up regularly.
> 
> I find, however, that basic searches don't always work well, or are too 
> slow.   Though, I admit I'm not 100% certain on how that process "works" 
> - for example, I use Thunderbird presently as my email client, and I'm 
> aware that Cyrus maintains its own *.index files in each directory in 
> the spool.
> 
> It would be wonderful to have a richer indexing method.  That would 
> likely include some integration into clients (beyond the scope of this 
> discussion).
> 
> Other programs I've run into include Zoë <http://zoe.nu/>, which is 
> highly specialized (ie: you must use its infrastructure), there are some 
> on freshmeat.net.
> 
> I wonder what others thoughts/experiences are on this area.  Or if there 
> are some other clever methods of enriching this process that I've 
> overlooked.

If you're looking for better SEARCH performance from Cyrus, you should 
create SQUAT indices for your mailboxes.  See squatter(8).

As far as the IMAP client is concerned, it can only do what RFC 3501 and 
any IMAP extensions allow.  For instance, there is no multiple mailbox 
SEARCH command, so the client has to do the work itself.

-- 
Kenneth Murchison
Systems Programmer
Project Cyrus Developer/Maintainer
Carnegie Mellon University