Archives

This is the archive for July 2007

More digital karma

Completed an extension to the solr search engine integrating my local lucene work with the out of a box search engine, and have called it local solr.

While Solr isn't exactly an enterprise level search engine, it's not a bad start. Am hoping that either I or someone folks I work with will be given the opportunity to extend it and make it enterprise oriented.

So what's wrong with it?
Well

  1. ) Requires timeout configuration
    - Perhaps an observer pattern design will provide this
  2. ) Needs to have a query engine, configurable and separate from presentation layer.
    - At the moment the xml,json, text output formats, also control the query format.
  3. ) Query engines need to be built to support both local and remote partitions.
    - Currently only 1 single partition can be used, in a single thread on 1 CPU, so obviously speed is reduced by number of documents dramatically.
  4. Both query and result processes need to have a chain, to allow you to plugin categorization, logging, normalization etc, and in the result chain, allow you to add dynamic fields from external sources.

It's a good chunk of work, but is possible with time and effort. Have to see if we'll be given those.