Percentiles(Kind=Keyed) order is not right

[shi-yuan]

Java API client version

8.5.3

Java version

17

Elasticsearch Version

8.5.3

Problem description

ElasticsearchClient elasticsearchClient = new ElasticsearchClient(getElasticsearchTransport(getRestClient()));
SearchRequest.Builder builder = new SearchRequest.Builder();
builder.index("elasticsearch-sql_test_index_account");
builder.query(Query.of(q -> q.matchAll(MatchAllQuery.of(e -> e))));
builder.aggregations("percentiles(age)", Aggregation.of(a -> a.percentiles(PercentilesAggregation.of(
        e -> e.percents(1.0, 5.0, 25.0, 50.0, 75.0, 95.0, 99.0).field("age")
))));
builder.size(0);
SearchRequest searchRequest = builder.build();
SearchResponse<Object> searchResponse = elasticsearchClient.search(searchRequest, Object.class);
System.out.println(searchResponse);

output:

SearchResponse: {
  "took": 4,
  "timed_out": false,
  "_shards": {
    "failed": 0,
    "successful": 1,
    "total": 1,
    "skipped": 0
  },
  "hits": {
    "total": {
      "relation": "eq",
      "value": 1000
    },
    "hits": [],
    "max_score": null
  },
  "aggregations": {
    "tdigest_percentiles#percentiles(age)": {
      "values": {
        "25.0": "25.0",
        "1.0": "20.0",
        "95.0": "39.0",
        "50.0": "31.0",
        "5.0": "21.0",
        "99.0": "40.0",
        "75.0": "35.0"
      }
    }
  }
}

but expected:

...
  "aggregations": {
    "tdigest_percentiles#percentiles(age)": {
      "values": {
        "1.0": "20.0",
        "5.0": "21.0",
        "25.0": "25.0",
        "50.0": "31.0",
        "75.0": "35.0",
        "95.0": "39.0",
        "99.0": "40.0"
      }
...

co.elastic.clients.elasticsearch._types.aggregations.Percentiles#buildPercentilesDeserializer use co.elastic.clients.json.JsonpDeserializerBase.StringMapDeserializer, but it's co.elastic.clients.json.JsonpDeserializerBase.StringMapDeserializer#deserialize method use HashMap,

I think, it should use LinkedHashMap to keep the order:

[sethmlarson]

JSON object keys are unordered per the JSON standard, so I believe the implementation is correct as-is. Why do you need the ordering of keys here if you can access each known key individually?

[shi-yuan]

I think, it is better to be consistent with the result of rest _search api