Question

Before asking this question I have googled for some time, but could not find any relevant information on this topic.

My problem is simple:

I have NHibernate criteria and projection and I’m trying to set DISTINCT ON(column)

My code for projection is following:

        criteria.SetProjection(
            Projections.ProjectionList()
            .Add(Projections.Distinct(Projections.Property("ID")))
            .Add(Projections.Property("A"))
            .Add(Projections.Property("B"))
         );

This generates following SQL (bit simplified):

SELECT DISTINCT ID, A, B FROM ABC ORDER BY A

But unfortunately the performance if this query is very-very poor.

I have optimized my SQL that runs much faster and looks like this:

SELECT DISTINCT ON (A) ID, A, B FROM ABC ORDER BY A

Is there anyway I can make NHibernate generate SQL that I have just shown?
Is this problem solvable with NHibernate’s dialects?

Looking forward to Your feedback!
Thank You very much!

Answer 1

This is not the most elegant solution, by any means but I’m wondering if you can accomplish what you need by using a “SQLProjection”? I’m a Java guy, so the code may not be exactly right here, but the following approach looks like it works for me using Postgres:

criteria.SetProjection(
    Projections.ProjectionList()
    .Add(Projections.SqlProjection("DISTINCT ON(A) ID"))
    .Add(Projections.Property("A"))
    .Add(Projections.Property("B"))
 );

Which generates the optimized SQL query you gave in your question. Again, this approach may not work if you’re doing something highly dynamic, but it is A way of tackling the problem.