Query: Translate Distinct to SQL when applied after custom projection

I looked up for this problem and didn’t find anything. Sorry if I’m making a dup or I messed up something.

This

context.Blogs
    .Select(t => new BlogDto
    {
        Url = t.Url
    })
    .Distinct()
    .ToList();

resulted in the following query:

SELECT [t].[Url]
FROM [Blogs] AS [t]

and I expected:

SELECT DISTINCT [t].[Url]
FROM [Blogs] AS [t]   

So, now I’m using the following workaround:

context.Blogs
    .Select(t => new
    {
        Url = t.Url
    })
    .Distinct()
    .Select(t => new BlogDto
    {
        Url = t.Url
    })
    .ToList();

which results in:

SELECT [t1].[Url]
FROM (
    SELECT DISTINCT [t0].[Url]
    FROM [Blogs] AS [t0]
) AS [t1]

Steps to reproduce

    public class BlogDto
    {
        public string Url { get; set; }
    }

    public class BlogContext : DbContext
    {
        public DbSet<Blog> Blogs { get; set; }

        protected override void OnConfiguring(DbContextOptionsBuilder optionsBuilder)
        {
            optionsBuilder.UseSqlServer(@"Data Source=.;Initial Catalog=Blogging;Integrated Security=True;MultipleActiveResultSets=True");
            //optionsBuilder.UseSqlServer(@"Server=(localdb)\mssqllocaldb;Database=CollectionsTest;Trusted_Connection=True;ConnectRetryCount=0");
        }
    }

    public class Program
    {
        private static void Main()
        {
            using (var context = new BlogContext())
            {
                var results = context.Blogs
                    .Select(t => new BlogDto
                    {
                        Url = t.Url
                    })
                    .Distinct()
                    .ToList();
            }
        }
    }

Just to be clear, the following works as expected:

context.Blogs
    .Select(t => new
    {
        Url = t.Url
    })
    .Distinct()
    .ToList();

produces:

SELECT DISTINCT [t].[Url]
FROM [Blogs] AS [t]

I have one simple question regarding terminology. When you say ‘projection’ (‘projecting’ to some object different from the model), does that include anonymous objects? If so, the title of this issue should be “Projecting to a specific object(class) with Distinct is not working as expected”

Thank you!

Further technical details

EF Core version: 1.1.0 Database Provider: Microsoft.EntityFrameworkCore.SqlServer Operating system: Win 8.1 64bit IDE: Visual Studio 2015 Update 3

[divega]

EF Team Triage: We discussed this and we believe we should try to support this query and the translation to DISTINCT on the server.

[adzhiljano]

Thanks for the response! You are right, but when working with a DbContext, I personally think of Distinct(IQueryable) as a way to force a distinct to be executed on the server(query provider). I’m sure that a lot of people are used to this way of query construction.

Also, if you run the following using EF 6

context.Blogs
    .Select(t => new BlogDto
    {
        Url = t.Url
    })
    .Distinct()
    .ToList();

you will get the following translation:

SELECT 
    [Distinct1].[C1] AS [C1], 
    [Distinct1].[Url] AS [Url]
    FROM ( SELECT DISTINCT 
        [Extent1].[Url] AS [Url], 
        1 AS [C1]
        FROM [dbo].[Blogs] AS [Extent1]
    )  AS [Distinct1]

which is the same query as the one I mentioned in the workaround and again I think a lot of people are used to this mindset when constructing a query. It’s just confusing for me that somehow the server should be aware of comparing c# classes.

I saw the following

EF Core keeps the developer experience from EF6.x

at https://docs.microsoft.com/en-us/ef/efcore-and-ef6/ which is not true in this case. If you are transitioning from EF6 to EFCore you will have to refactor some of your queries including Distinct.

Even more, if you run:

context.Blogs
  .Distinct()
  .ToList();

you will get

SELECT DISTINCT [b].[BlogId], [b].[Url]
FROM [Blogs] AS [b]

and I didn’t find an override of the Equals method in the DbSet. Maybe I’m missing something conceptually or I’m just too used to EF6. Thanks, again!