Inconsistent behavior between FT.SEARCH and FT.AGGREGATE

[mfaulcon]

In my project, a pair of IQueryable queries are being sent through Redis.OM together that are not behaving as expected. My object is defined as below:

[Document(StorageType = StorageType.Json, Prefixes = [Prefix], IndexName = Prefix + ":idx")]
public class CachedUser : IEntity<Guid>
{
    public const string Prefix = "graph3:user";

    [Indexed(Sortable = true, IndexEmptyAndMissing = false)]
    [RedisIdField]
    public Guid UserId { get; set; }

    // redacted

    [Indexed(Sortable = true, IndexEmptyAndMissing = true)]
    public string? FirstName { get; set; }

    // redacted
}

I'm executing the GraphQL query below:

query users {
  users(where: { firstName: { startsWith: "Ma" } }) {
    totalCount
    items {
      userId
      firstName
      lastName
      email
    }
    pageInfo {
      hasNextPage
      hasPreviousPage
    }
  }
}

At runtime, the query is being translated into the two queries below:

FT.SEARCH graph3:user:idx "(-(ismissing(@FirstName)) (@FirstName:{Ma*}))"
  DIALECT 2
  LIMIT 0 11
  RETURN 12
  UserId AS UserId
  FirstName AS FirstName
  LastName AS LastName
  Email AS Email

and

FT.AGGREGATE graph3:user:idx "@FirstName:{null} (@FirstName:{Ma*})"
  GROUPBY 0
  REDUCE COUNT 0 AS COUNT

The search returns data as expected, but the aggregate returns the count of all documents, instead of only the ones that match the search criteria. If I add DIALECT 2 to the FT.AGGREGATE statement and change the query to match the FT.SEARCH statement, I get the output I expect.

I am running on
Redis 7.4.2
RediSearch 2.10.17
RedisJSON 2.8.8

[mfaulcon]

@slorello89, I don't mind trying to add the dialect use case to the aggregate code, but I wouldn't even know where to get started...

[slorello89]

Ahh yes, FT.AGGREGATE has slightly different query parsing logic, and when we added the null checks, we only added them to the Search commands when I implemented the Null checks - will need to add this.