Where Should I Concentrate Effort?

[Garrett Rooney <ro...@electricjellyfish.net>]

So I'm trying to plan out the next week or two of development work for 
Lucene4c, with the goal being getting it to the point where it can 
actually be used for something other than playing around.

I think there are three major steps remaining.

1) Flesh out searcher and hits interfaces, so applications can use 
something approaching the interface they'll eventually get.  Most of 
this is going to be a rather thin layer over the scorer code, since 
until the scorer can actually calculate scores it does little good to 
try to rank hits.

2) Build a query parser, so that applications don't need to manually 
build up query objects.  This will almost certainly be a minimalistic 
parser, but it'll be enough to start using it.

3) Actually make the scorers calculate scores, and then make the 
searcher use them to rank hits.

What do people think?  If I can get those knocked off would it be enough 
for people to actually start making use of Lucene4c for something?

-garrett

Re: Where Should I Concentrate Effort?

[Garrett Rooney <ro...@electricjellyfish.net>]

Garrett Rooney wrote:
> So I'm trying to plan out the next week or two of development work for 
> Lucene4c, with the goal being getting it to the point where it can 
> actually be used for something other than playing around.
> 
> I think there are three major steps remaining.

Just a quick update...

> 1) Flesh out searcher and hits interfaces, so applications can use 
> something approaching the interface they'll eventually get.  Most of 
> this is going to be a rather thin layer over the scorer code, since 
> until the scorer can actually calculate scores it does little good to 
> try to rank hits.

This is next on my hit list after finishing the boolean scorer work.

> 2) Build a query parser, so that applications don't need to manually 
> build up query objects.  This will almost certainly be a minimalistic 
> parser, but it'll be enough to start using it.

The basics of this are now working, although it needs a LOT more work.

> 3) Actually make the scorers calculate scores, and then make the 
> searcher use them to rank hits.

This is last on the list, because I figure it'll take me a while to 
actually figure out how this part of Java Lucene works ;-)

> What do people think?  If I can get those knocked off would it be enough 
> for people to actually start making use of Lucene4c for something?

I've also now got the command line client set up to start making use of 
these new "useful" parts of the API.  Right now the 'lcn search' command 
is implemented in terms of the lcn_scorer_t API, but as soon as I get 
the searcher and hits code in place it'll migrate to that.

-garrett