solrj for distributed search

[Yonik Seeley <yo...@apache.org>]

I've been hacking on SOLR-303 (distributed search), and I started to
write my own XML parsing code utilizing stax (streaming), when I
realized that the code had already been written (in SolrJ).

Should we use SolrJ for making and parsing the distributed search requests?
One downside is that SolrJ would need to be moved into the core, but I
was planning on migrating to HttpClient at some point anyway.

Thoughts?

-Yonik

Re: solrj for distributed search

[Ryan McKinley <ry...@gmail.com>]

> Thoughts?
> 

I like this idea.  The more we reuse stuff, the  better it should get. 
If/when we figure a faster way to transmit request/response data (JSON? 
RMI?), it would be good to have this cooked into distributed search easily.

 > One downside is that SolrJ would need to be moved into the core, but I

hymmm.  true.  It would make core depend on a specific client.  Clients 
would still not need to include jar files for 'core' stuff.  It would 
have a dependancile graph like:

  + common
    + solrj
      + core

I guess common and solrj could be a single jar, but it may be useful to 
keep them separated.

ryan