[sword-devel] fast searching support

Joachim Ansorg sword-devel@crosswire.org
Sun, 20 Aug 2000 12:53:04 +0000


Hi Troy!

> > 1 second without indexing? What do you have for a expensive PC ? ;-)
> > My PC (AMD K6-2 400@300, 128MB Ram) needs some seonds!
> >
> :)  I was on my box at the office.  It is kindof a beefy machine.  I
>
> think it's a 600 with 512megs of memory running a 2.2.16 i686 kernel.

Nice thing.

> > BibleTime does now work with the indexing mechanism, it's not so fast as
> > cheatah but much more fast than before.

> I wonder why cheatah would be faster.  Are you sure that you are using
> indexing?  The search for "God love world" should be instantanious.

I'm using indexing. IMHO the slower speed is related to our object tree 
model. We create a a list of found items. Each found key has an own object. 
Creating it takes some time. I's a thing we should improve.

> > Often used take very lomg.
>
> Large result sets?

Yes.

> > BTW, Sword does behace strange now i you search the GerBen module for
> > "God love world", it shows some thousand 1. Mose 0:1 and some Revelation
> > entries (over 10000 found entries).
>
> Did you index this module?  Or is it running the traditional search?

If I use "Case sensitive" search it works correctly, but If I use 
"non-case-sensitive" search so the index-stuff is enabled it doesn't work. 
You can test it in cheatah.
BTW, the GerBen module does only have a New testament, the old is empty. 
Maybe this is causing the problem?

> 	-Troy.