Question

I’m wondering whether I should create extension methods that apply on the object level or whether they should be located at a lower point in the class hierarchy. What I mean is something along the lines of:

public static string SafeToString(this Object o) {
    if (o == null || o is System.DBNull)
        return "";
    else {
        if (o is string)
            return (string)o;
        else
            return "";
    }
}

public static int SafeToInt(this Object o) {
    if (o == null || o is System.DBNull)
        return 0;
    else {
        if (o.IsNumeric())
            return Convert.ToInt32(o);
        else
            return 0;
    }
}
//same for double.. etc

I wrote those methods since I have to deal a lot with database data (From the OleDbDataReader) that can be null (shouldn’t, though) since the underlying database is unfortunately very liberal with columns that may be null. And to make my life a little easier, I came up with those extension methods.

What I’d like to know is whether this is good style, acceptable style or bad style. I kinda have my worries about it since it kinda “pollutes” the Object-class.

Thank you in advance & Best Regards 🙂

Christian

P.S. I didn’t tag it as “subjective” intentionally.

Answer 1

No, that is not good practice. You want to apply extension methods at the lowest possible point. I believe there a time and a place for (almost) everything, but extension methods System.Object would almost never be appropriate. You should be able to apply extension methods such as this much further down the inheritance stack. Otherwise it will clutter your intellisense and probably end up being used/depended-upon incorrectly by other developers.

However, extension methods for data objects for dealing with Null values is a very good use of extension methods. Consider putting them right on you OleDbDataReader. I have a generic extension method called ValueOrDefault that . . . well, I’ll just show it to you:

<Extension()> _
Public Function ValueOrDefault(Of T)(ByVal r As DataRow, ByVal fieldName As String) As T
    If r.IsNull(fieldName) Then
        If GetType(T) Is GetType(String) Then
            Return CType(CType("", Object), T)
        Else
            Return Nothing
        End If
    Else
        Return CType(r.Item(fieldName), T)
    End If
End Function

That’s VB, but you get the picture. This sucker saves me a ton of time and really makes for clean code when reading out of a datarow. You are on the right track, but your sense of spell is correct: you have the extension methods too high.

Putting the extension methods into a separate namespace is better than nothing (this is a perfectly valid use of namespaces; Linq uses this), but you shouldn’t have to. To make these methods apply to various db objects, apply the extension methods to IDataRecord.