So I'm just wondering if the docs can be totally trusted as a complete source of truth for all possible field mapping properties on a data type.
I'm not much of a Java dev, but if pointed in the right direction I could probably grok the source. I've noticed the .NET client has slightly different fields than the docs, some of which are deprecated, some of which just don't seem to be mentioned anywhere.
I ask so that I can make sure I'm supporting mapping properly in my Rust client.
Sure, so for example numeric_resolution on date, which isn't in the docs (I think because it's deprecated?), but is in the .NET client (for back-compat?).
And fielddata (in the pre 5.x format) isn't listed on a bunch of fields in the docs (to keep the docs concise?), but is in the .NET client.
One last one is ignore_malformed on geo_point which is in the docs, but not in the .NET client (although it does have validate, which I think might have been changed to ignore_malformed for consistency at some point?).
I'm sure there are perfectly sane reasons for differences, but it'd be great if there was somewhere I could go with everything that is current for the upcoming 5.x I could use as a base.
I've gone spelunking in the Java source a few times, but couldn't be totally sure of what I was looking at
The master branch and the resulting master documentation should be considered a work in progress at this time; it's being actively worked on in line with changes happening with the 5.x alpha releases of Elasticsearch.