pysolr is pretty similar to the current code. It does add some more bits and adds API's for Solr Core.
sunburnt adds a powerful more natural query language, which is much better for querying Solr directly. It's similar to SQLAlchemy or jQuery.
I think the sunburnt approach is more compelling in the mid-term. The whole process of emulating the ZCatalog API's via Solr and making everything transparent is pretty painful. And it's pretty hard to make use of any of the advanced Solr features in this way. Facets kind-of work, but once you get into spelling corrections, more-like-this or other features it all gets quite painful.
By now there's two well maintained Python Solr libraries out there. We should look at them and see if we can use either of them:
https://github.com/toastdriven/pysolr/ http://packages.python.org/sunburnt/
pysolr is pretty similar to the current code. It does add some more bits and adds API's for Solr Core.
sunburnt adds a powerful more natural query language, which is much better for querying Solr directly. It's similar to SQLAlchemy or jQuery.
I think the sunburnt approach is more compelling in the mid-term. The whole process of emulating the ZCatalog API's via Solr and making everything transparent is pretty painful. And it's pretty hard to make use of any of the advanced Solr features in this way. Facets kind-of work, but once you get into spelling corrections, more-like-this or other features it all gets quite painful.