Elassandra Support Documentation says to use 'string' and 'index: "analyzed"' which is not available in newer versions of Elassandra

[AlexisWilke]

In your Elassandra support page, you have an example of mapping that uses string as a type and uses "analyzed" for the index field:

es_index_mapping: {
  discover: '.*',
  properties: {
      "name" : {
          "type" : "string",      // <- this is now wrong, should be `text` for this example
          "index" : "analyzed"    // <- this is now wrong, you may use `index: true`, although type `text` is probably enough
      }
  }
}

Newer versions of Elassandra changed string to either text (searchable, equivalent to string + index: "analyzed") or keyword. So we should probably change that.

Further, the new system will fail with an error saying that there is a type mismatch between the given type (text) and the inferred type (list<text>), which is due to the fact that by default fields are viewed as collections. To palliate you need to use the CQL collection parameter like this:

  properties: {
      "name" : {
          "type" : "text",
          "index" : true,
          "cql_collection" : "singleton"
      }
  }

This is probably since Elassandra version 5.x, for sure, 6.x requires such.

Finally, I had a problem with the concept of "To overwrite some properties, you could do the following:" — somehow that does NOT work, at least not for for text fields. So what I had to do with Elassandra 6.2.3.7 is add every single type definition manually and not use the discover: ... at all. I will probably be testing some more to see whether I could use a different pattern than ".*" to include fields that can automatically be generated and use specific field definitions for the rest. But overwrite is apparently the wrong word with the latest version, unfortunately. I'll post a bug report for that problem in Elassandra and see what they have to say about this. It may also be a problem in ElasticSearch itself.