Abstract object not mapped correctly in Elasticsearch using Nest 7.0.0-alpha1 - elasticsearch

Question

I am using NEST (.NET 4.8) to import my data, and I have a problem getting the mapping to work in NEST 7.0.0-alpha1.
I have the following class structure:
class LinkActor
{
public Actor Actor { get; set; }
}
abstract class Actor
{
public string Description { get; set; }
}
class Person : Actor
{
public string Name { get; set; }
}
I connect to Elasticsearch this way:
var connectionSettings = new ConnectionSettings(new Uri(connection));
connectionSettings.DefaultIndex(indexName);
var client = new ElasticClient(connectionSettings);
The actual data looks like this:
var personActor = new Person
{
Description = "Description",
Name = "Name"
};
var linkActor = new LinkActor
{
Actor = personActor
};
And the data is indexed like this:
result = client.IndexDocument(linkActor);
Using NEST 6.6 I am getting the following data in Elasticsearch 6.5.2:
"actor": {
"name": "Name",
"description": "Description"
}
However when using NEST 7.0.0-alpha1 I get the following data in Elasticsearch 7.0.0:
"actor": {
"description": "Description"
}
So the data from the concrete class is missing. I am obviously missing / not understanding some new mapping feature, but my attempts with AutoMap has failed:
client.Map<(attempt with each of the above classes)>(m => m.AutoMap());
Is is still possible to map the data from the concrete class in NEST 7.0.0-alpha1?

Answer 1

I found a workaround using the NEST.JsonNetSerializer (remember to install this), which allows me to pass a JObject directly:
Connect to Elasticsearch using a pool so you can add the JsonNetSerializer.Default:
var pool = new SingleNodeConnectionPool(new Uri(connection));
var connectionSettings = new ConnectionSettings(pool, JsonNetSerializer.Default);
connectionSettings.DefaultIndex(indexName);
var client = new ElasticClient(connectionSettings);
Convert the linkActor object from above to a JObject (JsonSerializerSettings omitted for clarity, add them to get CamelCasing):
var linkActorSerialized = JsonConvert.SerializeObject(linkActor);
var linkActorJObject = JObject.Parse(linkActorSerialized);
result = client.IndexDocument(linkActorJObject);
This gives the desired result:
"actor": {
"name": "Name",
"description": "Description"
}
It is a workaround, hopefully someone will be able to explain the mapping in the question.