scaled_float has precision problem

[donvanf]

Describe the feature:

Elasticsearch version (bin/elasticsearch --version):
5.0.1, Build: 080bb47/2016-11-11T22:08:49.812Z

Plugins installed: []

JVM version (java -version):
1.8.0_131
OS version (uname -a if on a Unix-like system):
CentOS release 6.5
Description of the problem including expected versus actual behavior:
Results from inner_hits named "a" should be returned only if my_parent's key equal 3.
Steps to reproduce:

Please include a minimal but complete recreation of the problem, including
(e.g.) index creation, mappings, settings, query etc. The easier you make for
us to reproduce it, the more likely that somebody will take the time to look at it.

PUT test
{
  "mappings": {
    "type": {
      "properties": {
        "scaled_float_field": {
          "type": "scaled_float",
          "scaling_factor": 100
        }
      }
    }
  }
}
POST _bulk
{"update": {"_index": "test","_type": "type","_id": "1"}}
{"doc": {"scaled_float_field": 0.1},"doc_as_upsert": true}
POST test/_search
{
  "query": {
    "range": {
      "scaled_float_field": {
        "lt": 0.1
      }
    }
  }
}

result

{
  "took": 1,
  "timed_out": false,
  "_shards": {
    "total": 5,
    "successful": 5,
    "failed": 0
  },
  "hits": {
    "total": 1,
    "max_score": 1,
    "hits": [
      {
        "_index": "test",
        "_type": "type",
        "_id": "1",
        "_score": 1,
        "_source": {
          "scaled_float_field": 0.1
        }
      }
    ]
  }
}

As Document mentions here, scaled_float is stored as long type. Why it also has precision problem?

[jpountz]

This is a bug indeed. Range queries are supposed to work as if the field was mapped as a double field that only contains rounded values.

[jpountz]

Even though this is considered a bug, I'd like to point out that relying on equality on doubles may be a source of problems for other reasons, especially since 0.1 cannot be represented accurately as a double.