Hacker News Reader

HNHacker News

Top New Best Ask Show Jobs

Comment by morganherlocker | Hacker News Reader

Parent Full thread

morganherlocker·

Might not matter, but they do not follow the geojson spec for spatial storage.

View on HN

Sure, ES supports lat/lon as properties, strings, geohash and geojson:

http://www.elasticsearch.org/guide/en/elasticsearch/referenc...

I could be totally wrong, but the docs you linked to do not actually conform to the geojson spec. It is geographic and it is json, but not valid geojson. The part where it says:

> Format in [lon, lat], note, the order of lon/lat here in order to conform with GeoJSON.

.. the data example below is not actually geojson. See the spec:

http://geojson.org/geojson-spec.html

I think the documentation is not clear here. ElasticSearch has an internal GeoPoint type, which can be read from any kind of JSON document. One of the possible notations is the GeoJSON coordinate notation.

Elasticsearch can map any kind of JSON, so you can, without problems, write a mapping for proper GeoJSON points. (map "type" as unanalyzed string, map "coordinate" as GeoPoint). Arrays of values are generally supported in ES.

The biggest problem is that Elasticsearch probably does not provide all kinds of queries you'd like if you are working with complex shapes. Basically, only distance and simple location queries with polygons are supported.

Thanks for the explanation. Sounds like ES has equal or slightly more support for spatial stuff than MongoDB then (about as much as most people need).

Well, thanks for pointing it out. That's a good opportunity for improval :).

Reply on news.ycombinator.com