Question

Well, I think trying to understand someone elses SQL can be trying enough but trying to convert said SQL to Entity Framework is another level of pain…

I have this SQL:

SELECT 
    MAX(IntegerColumn) - MIN(IntegerColumn) 
    / 
    DateDiff(day, MIN(DateColumn) , MAX(DateColumn)) 
    * 
    DateDiff(day, MAX(DateColumn) , @TargetDate) 
    + 
    MAX(IntegerColumn) 
    AS Calculation      
FROM MyTable 
WHERE TargetId = @TargetId

Forgetting operator precedence, so far, I have come up with the following LINQ to Entities, which I am regarding with extreme suspicion:

var calculation = 
    (from f in this.ObjectContext.MyTable 
        where f.TargetId == targetId
        group f by f.IntegerColumn into o
        let maxIntegerColumn = o.Max(x => x.IntegerColumn)
        let minIntegerColumn = o.Min(x => x.IntegerColumn)
        let maxDate = (from t in o select t.DateColumn).Max()
        let minDate = (from t in o select t.DateColumn).Min()
        select new
        {
            Result = (maxIntegerColumn - minIntegerColumn) /
            ((SqlFunctions.DateDiff("day", minDate, maxDate) *
            SqlFunctions.DateDiff("day", maxDate, targetDate)) + maxIntegerColumn)
        }).FirstOrDefault();

if (calculation != null)
{
    if (calculation.Result != null)
    {
        return calculation.Result.ToString();
    }
}

That just looks wrong, wrong, wrong and is the result of blindly following too many blog posts.

Is this in the right direction? How can I simplify what’s going on?

Answer 1

If it works, go with it. You can reformat and refactor your LINQ to make it look prettier but it appears as though it’s doing the right thing.

In this case, I’d probably write a stored procedure – you’re performing awkward calculations that need to be done in the database, so just write them in the database and then execute them from your code.