[Xapian-discuss] More than one Index?

[James Aylett]

On Mon, Jan 30, 2006 at 01:15:41PM +0000, John Wards wrote:

> Ah so my scriptindex input file would contain:
> 
> description index=XDESC
> 
> And if I was searching for "character" I would have to add XDESC to it
> so the search would be "XDESCcharacter"

Yes.

> Now in terms of database size what difference would this make to the
> database? The description is at least 5 times the length of the address
> index and I have 650,000 documents...would I have to times the database
> by 5 to get an idea? Which is quite scary as the database is 1.1 gig in
> size at the moment.

It depends on how much of the descriptions are standard words, but
your positional and postlist tables will scale up like that (modulo
whatever compression is available - I think flint compresses
postlists, but can't remember the details).

If you're worried about database sizes, you could drop positional data
unless you need it (phrase queries, for instance).

How big is your source data? That's about 18k per document, which is
fairly large unless I misunderstand your application...

J

-- 
/--------------------------------------------------------------------------\
  James Aylett                                                  xapian.org
  james at tartarus.org                               uncertaintydivision.org