Description
In #42035 @jamesdorfman added a neat new kind of aggregations which performs similarly to a regular histogram but instead to specifying the width of each bucket you specify how many buckets you'd like, kind of like the auto_date_histogram. Unlike auto_date_histogram, variable_width_histogram's buckets have, well, variable width. The particulars of its algorithm mean that the buckets won't necessarily have the same number of documents in them. Instead it tries to preserve the shape of the data. When it has to merge buckets together it merges the buckets that are closest together already.
This is an interesting and new aggregation, and, frankly, we're not sure that we've really thought through the interface all the way. So we'd like some feedback on it and some time to be able to make breaking changes to its request and/or response. This, we're marking it experimental. Our plan to remove experimental is to have a few folks take a look at it and comment.