Question

Sample in C# and VB.NET are OK.

I have a table “People” with the following columns:

-FullName (nvarchar not null)
-DOB (datetime null)

I want to write a LINQ to SQL to group the people by age, like following result:

Age 19: 4 ppl

Age 20: 5 ppl

Age 21: 6 ppl

and so on…

Here’s my try:

Dim query = From ppl In db.People _
         Select New With {.Age = DateTime.Now.Year - CDate(ppl.DOB).Year, .CountAge = ppl.Count}

Notice that there are no DOB record for some people in the tables, so these shall not be included. The DOB column has record like this 1982-10-24 10:12:45 AM because it’s a DateTime column.

Answer 1

I would rather:

var result = db.People
  .Where(p => p.DOB != null) // exclude null DOB's from stats
  .GroupBy(p => p.DOB.Value.Year)
  .Select(g => new { Age = DateTime.Now.Year - g.Key, Count = g.Count() });

We are grouping against year of birth here. This should translate to GROUP BY YEAR(dob) in SQL which could have slightly better performance or optimization compared to GROUP BY YEAR(GETDATE()) - YEAR(dob). It’s almost impossible to map row expressions into indexes but certain constructs like YEAR() alone could be optimized to use a datetime index partially. I make many assumptions here though, like that datetime structures start with year, SQL server cares about YEAR(x) for optimizations etc. It’s still good to keep in mind such situations when constructing a LINQ query.