To make cloudelastic useful to developers they need to know what is actually in there. Document the schema somewhere (wikitech? mw.org?). This schema will also be useful documentation for the CirrusSearch dumps.
Information that should probably be included about each property:
- What is this property?
- Which ways is it analyzed?
- How stable is it?
Likely we also need some documentation at a higher level than the individual properties. The various analysis chain variants that are applied (keyword, near_match, near_match_asciifolding, plain, prefix, prefix_asciifolding, trigram, etc) will need to be documented as well.