Hstore as part of a larger search index is really handy. We're using it at DocumentCloud to allow reporters to tag and search through their primary source docum

Hstore as part of a larger search index is really handy. We're using it at DocumentCloud to allow reporters to tag and search through their primary source documents by adding their own arbitrary metadata. http://blog.documentcloud.org/blog/2011/05/arbitrary-metadata/ http://blog.documentcloud.org/blog/2011/05/arbitrary-metadat... ... when I search for "company:XTO" (or "company:*" for that matter), it can be serialized into an hstore query like so: https://github.com/documentcloud/documentcloud/blob/master/lib/dc/search/query.rb#L311-330 https://github.com/documentcloud/documentcloud/blob/master/l... ... and then be used to build apps with hstore queries on top of the API, like this: http://www.nytimes.com/interactive/2011/12/02/us/oil-and-gas-leases.html http://www.nytimes.com/interactive/2011/12/02/us/oil-and-gas... ... all without ever having to change the database schema, or indexes.