views:

73

answers:

1

I use Nullable<T> types quite a lot when loading values into classes and structs when I need them to be nullable, such as loading a nullable value from a database (as the example below).

Consider this snippet of code:

public class InfoObject
{
    public int? UserID { get; set; }
}

// Load the ID into an SqlInt32
SqlInt32 userID = reader.GetSqlInt32(reader.GetOrdinal("intUserID"));

When I need to load a value into a nullable property, sometimes I do this:

infoObject.UserID = userID.IsNull ? (int?)null : userID.Value;

Sometimes I do this instead:

infoObject.UserID = userID.IsNull ? new int?() : userID.Value;

Although they achieve the same result, I'd like to see if anyone knows which is better to use between (int?)null and new int?() with regards to performance, smallest IL code, best practices etc.?

Generally I've been favouring the new int?() version of the code above, but I'm not sure whether casting (int?)null is quicker for the compiler to resolve than new int?().

Cheers!

+5  A: 

They will generate the same IL. Use whichever you find easier to read.

When you talk about something being "quicker for the compiler to resolve" are you talking about micro-optimising your build times? I have no idea which would be faster there, but I doubt it will make a reliably-measurable difference for any significant body of code.

Personally I use null whenever I can do so with no ambiguity; I think I still generally prefer the cast to calling new, but it makes no real difference.

Jon Skeet
I agree with you. `(int?)null` looks far better than `new int?()` especially when used within in-line conditional statements like my example shows. Regarding your question: you answered it when you said the IL is the same. Build times aren't currently an issue (unless I use TFSBuild, but that's a whole different story) so so-long as the IL matches, I'm happy :)
Codesleuth