[fpc-pascal] ioda joda fulltext search not working

[Michael Van Canneyt]

On Fri, 11 Nov 2011, noreply at z505.com wrote:

>>
>>
>> On Wed, 9 Nov 2011, Graeme Geldenhuys wrote:
>>
>>> On 9 November 2011 10:40,  <michael.vancanneyt at .....> wrote:
>>>> It definitely works, I also use it in a commercial project to implement
>>>> full-text-search on a firebird database.
>>>
>>>
>>> Now that would be pretty awesome! :)
>>
>> It is.
>> When I showed it to the customers, they were very impressed :-)
>> (considering their question was 'we want to do like google search but
>> on the database')
>>
>
> I was reading recently that MySQL has fulltext on InnoDB built in.. don't
> know much about it but I came across that recently while googling.
> Firebird has no fulltext built in to it I guess, eh?

No, it has not.

And if you look at the mysql implementation of fulltext, you'll know why.
It's not very efficient.

>
> Also does the Ioda Joda have positions of your search words found? Or it
> just weighs the results and throws away any positions of the words. I.e.
> if the first instance of "searchterm" was found, does it record that
> position, or just throwaway position info?  I was thinking of using it
> like google too, but having the positions of the search terms in the
> document would help to give a preview of the text like google. Or is there
> some other way joda does this? You could just parse the entire document
> without using Joda for that (separate utility) and use joda for the
> weighting only, but I wondered if joda had something built in to memorize
> where the search terms were located in the document.

I can't speak about joda features, but the solution I worked on with Darius saves 
the position and up to 500 bytes of 'context' (in the default configuration), 
so it can show a preview.

Michael.