> -----Original Message-----
> From: Brian Goetz [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, November 13, 2001 8:58 AM
> To: Lucene Users List
> Cc: [EMAIL PROTECTED]
> Subject: Re: Memory Usage?
>
>
> > Since this is changing behavior that people are depending
> on, what about
> > creating a new QueryParser called QueryParserSafe that
> excludes option.
> > I don't like the idea of removing functionality with no backward
> > compatibility.
>
> I knew this was coming.
>
> I'm sorry, but I have to laugh just a little bit. The new query
> parser has only existed for less than two months -- and people have
> built empires based on it? I'm perfectly willing to debate whether
> its a good idea or not to remove the wildcard match syntax from the
> query parser, but I think the "backward compatibility" argument is one
> of the less compelling arguments against doing so. Bear in mind that
> no one is suggesting removing the functionality from the core -- just
> restricting its use to programmatically generated queries. A strong
> argument can be made for not exposing the "don't try this at home"
> behavior through an interface that is bound to be used by naive
> end-users.
>
How about this:
"You must have at least four non-wildcard characters in a word before
you introduce a wildcard." (source:
http://www.northernlight.com/docs/search_help_optimize.html)
I think the best approach would be to have a parameter (of query
parser?, of indexsearcher?) to set the minimal non wild-char characters
before any wildchar.
peter