INFO: Variable Coercion Rules in Visual Basic

MORE INFORMATION

Coercion between numeric types is pretty straightforward. Smaller types (integer) coerced to larger types (double) simply have their value assigned. Larger types (double) coerced to smaller types (integer) have their value assigned, but generate an Overflow error if the smaller type cannot contain the larger type's value. Other coercion rules are not as obvious.

NOTE: Any type can be implicitly coerced to any other type (excluding objects).

   Source Type    Coerced to      Apply this rule
   --------------------------------------------------------------

   Integer        Boolean         0=False, non-zero=True

   Boolean        Byte            False=0, True=255

   Boolean        any numeric     False=0, True=-1
                  (except Byte)

   String         Date            String is analyzed for mm/dd/yy,
                                  and so on

   Date           numeric type    Coerce to Double and use
                                  DateSerial(Double)

   numeric type   Date            Use number as serial date, check
                                  valid date range

   numeric type   Byte            Error if negative

   String         numeric type    Strings are treated as a Double
                                  when they need to represent a
                                  number
				

Some of these rules allow for simplified syntax when writing your code. For example:

   ReturnCode% = FunctionThatReturnsAnInteger()
   If (ReturnCode%) Then ...
				

This piece of code takes advantage of the implicit Integer to Boolean coercion and correctly evaluates the expression in the If statement.

Note on Strings as Numbers

Treating Strings as Doubles when they need to represent a number gives the String the maximum possible range and nearly the best possible precision (only Currency can have more precision, at a sacrifice in range). Because these coercions are generated by the compiler, the rule must be decided up front without regard to the actual content of the string.

Note that "treating as Double" is not quite the same as "coercing to Double." For example, addition and subtraction operators treat Currency as a preferable type to Double. Currency plus String will be treated as Currency plus Double, which would use Currency addition. Thus the string will be coerced directly to Currency.

Why Use Coercion Rules?

The largest reason is performance. Hard-coded coercion rules make Visual Basic version 4.0 faster and more efficient and provide backward compatibility to previous versions of Visual Basic. One of the big speed advantages comes because Visual Basic version 4.0 now knows the data type of control properties. Knowing the type of properties provides trememdous performance advantages. Without it, for example, setting a property required Basic to package the value in a Variant, and the control to unpack and coerce it to the right type. Now the control knows it is receiving the correct specific type, and Basic does a direct coercion to that type without ever involving the overhead of a Variant.

Unexpected Results

The following code pieces show a few common scenarios where errors can be generated by Visual Basic if types and coercion rules are not carefully considered:

   Dim I As Integer, J As Integer
   Dim L As Long, M As Long
   Dim S As String

   I = 32767
   L = 32767
   M = 1
   S = "Hello World!"

   J = I + 1
   ' Overflow. Integer upper limit = 32767. No coercion applied.

   J = I + M
   ' Overflow. I+M is coerced to a Long and the value 32768 is generated,
   ' but when the assignment operator is resolved, this Long value is
   ' coerced back to Integer (J's type) and overflows.

   L = I + 1
   ' Overflow. Coercion is not applied until the assignment operator is
   ' resolved. I + 1 'Overflow's the temporary Integer variable created
   ' to resolve the plus operator.

   If I Then Print "I is True"
   If Not I Then Print "Not I is True"
   ' This prints both 'I is True and Not I is True', a logical
   ' contradiction. The value of I = 32767, which is non-zero and under
   ' the Integer to Boolean rule, is coerced to True. With 'Not I',
   ' the NOT operator is applied first, which produces -32768 which is
   ' also non-zero and, therefore, True.
				

The following scenario shows how implicit conversion can generate a variety of unexpected results. You can add this code to a new program and run it too see the output.

   Private Sub Form_Click()
      Call MySub(1, 23) '<-- Note: passing *numbers*!
   End Sub

   Private Sub MySub (a As String, b As String)
      Debug.Print "a = "; a, , TypeName(a)
      Debug.Print "b = "; b, , TypeName(b)
      Debug.Print "a + b = "; a + b, , TypeName(a + b)
      Debug.Print "a + b - 1 ="; a + b - 1, TypeName(a + b - 1)
      Debug.Print "1 - a + b ="; 1 - a + b, TypeName(1 - a + b)
      Debug.Print "a + b + 1 ="; a + b + 1, TypeName(a + b + 1)
      Debug.Print "1 + a + b ="; 1 + a + b, TypeName(1 + a + b)
      Debug.Print "(1 + b) / a ="; (1 + b) / a, TypeName((1 + b) / a)
   End Sub
				

The output is:

   a = 1             String
   b = 23            String
   a + b = 123       String
   a + b - 1 = 122   Double
   1 - a + b = 23    Double
   a + b + 1 = 124   Double
   1 + a + b = 25    Double
   (1 + b) / a = 24  Double