[Xapian-discuss] Emulating SQL JOIN in xapian

James Aylett james-xapian at tartarus.org
Mon Mar 2 10:17:48 GMT 2009


On Mon, Mar 02, 2009 at 07:27:25AM +0000, Olly Betts wrote:

> > > 1) Simply duplicate the recipient information on every payment record.
> > > This is possible as there aren't likely to be more than 20 payments
> > > per recipient and I have the disk space.  I can see this speeding
> > > everything up too.
> > 
> > This will work, and I believe is the most obvious way of attacking the
> > problem in Xapian, if you must do so.
> 
> I wonder if there's a way to achieve this with PostingSource and
> MatchSpy.

I've never used or even looked particularly hard at PostingSource, but
I'm guessing half of it (feeding the relevant recipient record for
every payment record) could be done providing there's a way of
determining the Document id of the recipient for each payment
Document.

Does that allow that pair of documents to contribute weight to the
same output document? I can't tell from the posting source docs. A
match decider could then get rid of recipient records as needed?

Or I may be misunderstanding what you're suggesting entirely.

J

-- 
  James Aylett

  talktorex.co.uk - xapian.org - uncertaintydivision.org



More information about the Xapian-discuss mailing list