this is interesting - could you talk a bit about your solr schema ? In a very naive way I would be simply storing it all as a indexable text blob. Did you do anything more, just so that you can avoid a database ?
Sure. Well, basically Solr is just a flat index, although you can store now mutliple-values in one field and even make joins. Here is the clou: Everything is preprocessed and stored in a mysql-db and solr just "caches" everything. Its some kind of the flattened mysql schema. So I have a combination of a relational db and all the benefits with the huge bonus of the speed and reliability of solr in between. I started of with just a MySQL db, but real-time scoring is really a pain in the "back" with mysql and I couldn't figure a nice way to do it. On the other hand, if you expand a story, you see that the comments/upvotes from the sources are fetched, because joins like that are no possible in solr (read: I couldn't find a way to do it).
So, all the information is preprocessed at index time and solr is basically a text-blob spread across multiple fields, yes.
