Entity framework: cross join throws OutOfMemoryException
I have a table with 1.5 million records in SQL Server 2008. It has a varchar column 'ReferenzNummer' that is indexed.
The following query executed in SQL Management Studio is up and running fast:
SELECT v1.Id, v2.Id FROM Vorpapier as v1 cross join Vorpapier as v2
WHERE v1.ReferenzNummer LIKE '7bd48e26-58d9-4c31-a755-a15500bce4c4'
AND v2.ReferenzNummer LIKE '7bd4%'
(I know the request doesn't make much sense, there will be more restrictions, but that's not important at the moment)
Now I would like to execute a query like this from Entity Framework 5.0, my LINQ looks like this:
var result = (from v1 in vorpapierRepository.DbSet
from v2 in vorpapierRepository.DbSet
where v1.ReferenzNummer == "7bd48e26-58d9-4c31-a755-a15500bce4c4" &&
v2.ReferenzNummer.StartsWith("7bd4")
select new { V1 = v1.Id, V2 = v2.Id })
.Take(10)
.ToList();
This tries to load the entire table into memory, which will throw an OutOfMemoryException after a while. I tried to move the WHERE parts with no success:
var result = (from v1 in vorpapierRepository.DbSet.Where(v => v.ReferenzNummer == "7bd48e26-58d9-4c31-a755-a15500bce4c4")
from v2 in vorpapierRepository.DbSet.Where(v => v.ReferenzNummer.StartsWith("7bd4"))
select new { V1 = v1.Id, V2 = v2.Id })
.Take(10)
.ToList();
Can Entity Framework be told to create a cross join statement like the one I wrote?
UPDATE 1
The generated EF SQL looks like this (for both queries)
SELECT [Extent1].[Id] AS [Id],
[Extent1].[VorpapierArtId] AS [VorpapierArtId],
[Extent1].[ReferenzNummer] AS [ReferenzNummer],
[Extent1].[IsImported] AS [IsImported],
[Extent1].[DwhVorpapierId] AS [DwhVorpapierId],
[Extent1].[Datenbasis_Id] AS [Datenbasis_Id]
FROM [dbo].[Vorpapier] AS [Extent1]
UPDATE 2
When I change the LINQ query and join the table directly on the DatenbasisIDd field (this is not exactly what I want, but it might work), EF creates a join:
var result = (from v1 in vorpapierRepository.DbSet
join v2 in vorpapierRepository.DbSet
on v1.DatenbasisId equals v2.DatenbasisId
where v1.ReferenzNummer == "7bd48e26-58d9-4c31-a755-a15500bce4c4" && v2.ReferenzNummer.StartsWith("7bd4")
select new { V1 = v1.Id, V2 = v2.Id })
.Take(10)
.ToList();
The resulting SQL query looks like this. It works and fast enough.
SELECT TOP (10) 1 AS [C1],
[Extent1].[Id] AS [Id],
[Extent2].[Id] AS [Id1]
FROM [dbo].[Vorpapier] AS [Extent1]
INNER JOIN [dbo].[Vorpapier] AS [Extent2]
ON ([Extent1].[Datenbasis_Id] = [Extent2].[Datenbasis_Id])
OR (([Extent1].[Datenbasis_Id] IS NULL)
AND ([Extent2].[Datenbasis_Id] IS NULL))
WHERE (N'7bd48e26-58d9-4c31-a755-a15500bce4c4' = [Extent1].[ReferenzNummer])
AND ([Extent2].[ReferenzNummer] LIKE N'7bd4%')
I still can't see why EF is not creating a cross join in the original request. Is it just not supported?
source to share
If you use a join in a linq expression, it will be passed back to SQL Server. Here are some examples of the join operator in linq: http://code.msdn.microsoft.com/LINQ-Join-Operators-dabef4e9
source to share