Elasticsearch 1.0 has been released: http://www.elasticsearch.org/blog/1-0-0-released/ Review what's changed and update our code to use the new-ness: http://www.elasticsearch.org/guide/en/elasticsearch/reference/1.x/breaking-changes.html Take note of deprecations: http://www.elasticsearch.org/guide/en/elasticsearch/reference/1.x/_deprecations.html
Also, upgrade to elasticutils 0.9 (currently unreleased and not on pypi) to pull in some required changes. This may be optional. EU supports 'match' instead of 'text' if we use it. EU 0.9 also switches to using the official elasticsearch python library which would be nice to use if we could.
For changing text queries to match queries which are supported by both ES 0.90.x and 1.0: https://github.com/mozilla/zamboni/commit/1346b8c
https://github.com/mozilla/zamboni/commit/0fab353 This commit changes the handling when querying a subset of fields to always return a list. An equivalent patch has also been submitted to elasticutils for review: https://github.com/mozilla/elasticutils/pull/206 Next stop: Updating how we do query time boosting: http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/mapping-boost-field.html#function-score-instead-of-boost
(In reply to Rob Hudson [:robhudson] from comment #3) > Next stop: Updating how we do query time boosting: > http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/ > mapping-boost-field.html#function-score-instead-of-boost Did some thoughts on this while at pycon: https://github.com/mozilla/elasticutils/issues/233 Of note, however, is that Elasticsearch (the company) is working on a higher level Python library for building queries. Once released we could switch to it over elasticutils. The developers of elasticutils are also considering a library on top of this new library to provide the ORM-like querying. Either way, unless this is urgent I'm proposing we wait for these new libraries. Lowering priority.
Closing tracker bug, we're on Elasticsearch 1.2.x now.