CODEWITHSUNDEEP
c#.netcastingnullabledbnull
Bull
Bull

Reputation: 701

Cast a null into something?

I had this interesting discussion today with a colleague. We were debating two pieces of code in C#.

Code Snippet 1: 

if(!reader.IsDBNull(2))
{
  long? variable1 = reader.GetInt64(2)
}

Code Snippet 2:

long variable1 = reader.IsDBNull(2) ? (long?) null : reader.GetInt64(2)

Question is: is it a good practice to cast null into a nullable long? Or would you rather use the traditional if statement to avoid casting null to nullable long.

Upvotes: 5

Views: 20704

Answers (5)

Daniel Liuzzi
Daniel Liuzzi

Reputation: 17147

In C# 7.1 you can use the more concise default literal:

var variable1 = reader.IsDBNull(2) ? default : reader.GetInt64(2);

Upvotes: 0

Sergey Berezovskiy
Sergey Berezovskiy

Reputation: 236208

I prefer not to cast null value (it looks odd to me):

long? variable1 = reader.IsDBNull(2) ? null : (long?)reader.GetInt64(2);

Another options:

long? variable1 = reader.IsDBNull(2) ? default : reader.GetInt64(2); // requires C# 7.1
long? variable1 = reader.IsDBNull(2) ? default(long?) : reader.GetInt64(2);
long? variable1 = reader.IsDBNull(2) ? (long?)null : reader.GetInt64(2);
long? variable1 = reader.IsDBNull(2) ? new Nullable<long>() : reader.GetInt64(2);
long? variable1 = reader.IsDBNull(2) ? new long?() : reader.GetInt64(2);
long? variable1 = reader.IsDBNull(2) ? null : new long?(reader.GetInt64(2));

It's just the matter of taste. I think first option is more readable, than others.

UPDATE: Consider also writing some extension methods to make your code more clear:

public static class DataReaderExtensions
{
    public static long? GetNullableInt64(this IDataReader reader, int index)
    {
        if (reader.IsDBNull(index))
            return null;

        return reader.GetInt64(index);
    }
}

In this case you don't use ternary operator (no casting to nullable), and reading values from reader looks more pretty:

long? variable1 = reader.GetNullableInt64(2);

Upvotes: 1

Tilak
Tilak

Reputation: 30698

Instead of 

(long?) null

use

default(long?)

I would refactor above code like

long? variable1 = reader.IsDBNull(2) ? default(long?) : reader.GetInt64(2)

Upvotes: 3

Wormbo
Wormbo

Reputation: 4992

The expressions (type?)null, default(type?) and new Nullable<type>() end up being compiled into the same opcodes:

        long? x = (long?)null;
        long? y = default(long?);
        long? z = new Nullable<long>();

is turned into:

    IL_0001: ldloca.s x
    IL_0003: initobj valuetype [mscorlib]System.Nullable`1<int64>
    IL_0009: ldloca.s y
    IL_000b: initobj valuetype [mscorlib]System.Nullable`1<int64>
    IL_0011: ldloca.s z
    IL_0013: initobj valuetype [mscorlib]System.Nullable`1<int64>

In other words, if you are working with nullable types, you are free to use whichever version you like best. Note however, that you should try to avoid arithmetics with nullable types. If you want to return a nullable value from a conditional expression, both possible results must be nullable if one of them can be null. Any other way could cause an exception in that case.

Upvotes: 15

undefined
undefined

Reputation: 1364

Snippet 2 is worth in my case, as in case of null you're gonna get 0, which is a completely valid value for long

Upvotes: 0

Related Questions