Hi
----- Original Message -----
From: Rafal Krzewski <[EMAIL PROTECTED]> Sent: Monday, May 29, 2000 3:19
AM
> > I would love to be able to do this, but if there is *anything* that is
> > specific to Lucene in there, I can't allow it because of license
> > issues...Lucene is unfortunately GPL.
> I hope this can be sorted out somehow. Someone asked the author
> of Lucene about licensing issues on the mailing list a month ago,
> but no aswer arrived.
> (see http://www.geocrawler.com/archives/3/2624/2000/4/0/3646428/)
>
> Would it be possible to make a generic search engine interface in
> Turbine, and write a wrapper around Lucene, for plugging it in?
> The wrapper would be covered by GPL, and Turbine wouldn't be
> influenced. Am I thinking right?
>
The two interface I posted on the previous message could (theoretically) be
wrapped around any search engine. The original plan was to give uniform
access to a wide range of search engines, but I realize that it can also
solve licence problems :-).
At the moment things look like this: There are 2 interfaces to define the
access layer to search engines. Classes that whish to add search
capabilities to turbine should implement these interfaces with their own
specialized code - I have done this for lucene. Therefore the interfaces
are implementation independent (and has *nothing* to do with Lucene or any
other search engine for that matter :-), but the Lucene implementation of
the interfaces requires lucene.jar.
If we cannot add the lucene.jar (which will probably be the case) to the
Turbine project I could maybe do a default implementation of a Boyer-Moore
algorithm [there is a pretty good discussion in Michael Abrash's Graphics
Programming Black Book, unfortunately everything in C and ASM :-(] that
simply search through text files as a default implementation for the search
interfaces. Users could use this in design/testing phases and then move to
a real search engine later on.
The lucene wrapper implementations could then be a separate download
(wrappers *only*) or people could get it from me (Maybe put it on our
website). The Excalibur Retrievalware implementation I'm working on at the
moment would need to work like this anyway, for it is a commercial product.
If licensing is still an issue we could only go with the Boyer-Moore
solution without any mention of other search engines although I feel that
this is not the best solution.
Any thoughts on this?
Regards
Leon
------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Problems?: [EMAIL PROTECTED]
