Simplification of Count(*) with multiple group by can generate wrong result

[smitpatel]

ss.Set<Order>().GroupBy(e => e.CustomerID)
                        .Select(g => new
                        {
                            g.Key,
                            A = ss.Set<Employee>().Where(e => e.City == "Seattle").GroupBy(e => e.City)
                                    .Select(g2 => g2.Count() + g.Min(e => e.OrderID))
                                    .OrderBy(e => 1)
                                    .FirstOrDefault()
                        })

Generates

SELECT [o].[CustomerID] AS [Key], COALESCE((
    SELECT TOP(1) COUNT(*) + MIN([o].[OrderID])
    FROM [Employees] AS [e]
    WHERE [e].[City] = N'Seattle'
    GROUP BY [e].[City]
    ORDER BY (SELECT 1)), 0) AS [A]
FROM [Orders] AS [o]
GROUP BY [o].[CustomerID]

Which works correctly.
But if the g.Min(e => e.OrderID) is converted to g.Count() then generate the SQL like this

SELECT [o].[CustomerID] AS [Key], COALESCE((
    SELECT TOP(1) COUNT(*) + COUNT(*)
    FROM [Employees] AS [e]
    WHERE [e].[City] = N'Seattle'
    GROUP BY [e].[City]
    ORDER BY (SELECT 1)), 0) AS [A]
FROM [Orders] AS [o]
GROUP BY [o].[CustomerID]

Now all of sudden the aggregate which is supposed to count records on outer grouping is computing that on inner one producing wrong results.

[smitpatel]

We should block query patterns like this. Basically when the COUNT(*) is used if there are multiple GroupBys. The subquery (non-lifted translation would also cause incorrect results. Users can specify specific column to count on.