Code-a-holic: [VB6] MAXDOUBLE, MINDOUBLE, +INFINITY, -INFINITY and NaN

Tuesday, May 19, 2009

[VB6] MAXDOUBLE, MINDOUBLE, +INFINITY, -INFINITY and NaN

VB6 doesn't seem to be able to specify that a Double contains MAXDOUBLE, MINDOUBLE, +INFINITY, -INFINITY or NaN. (I could be wrong on that as there may be some undocumented feature hiding in there somewhere.)

Anyway, I came up with a way of storing these values into Doubles so that they could be used in a Complex Numbers library I've been writing.

Essentially, I create an 8 byte array, load it up with the relevant values, and then, using API calls and the VarPtr function, store the contents of the array into the storage used by the Double.

Here's the code. First the declarations.

Public MaxDouble As Double
Public MinDouble As Double
Public NegInfinity As Double
Public PosInfinity As Double
Public QuietNAN As Double

Private Declare Sub CopyMemory Lib "kernel32" _
    Alias "RtlMoveMemory" (Destination As Any, _
    source As Any, ByVal Length As Long)
Private Declare Function GetProcessHeap Lib "kernel32" () As Long
Private Declare Function HeapAlloc Lib "kernel32" _
    (ByVal hHeap As Long, ByVal dwFlags As Long, _
     ByVal dwBytes As Long) As Long
Private Declare Function HeapFree Lib "kernel32" _
    (ByVal hHeap As Long, ByVal dwFlags As Long, lpMem As Any) As Long
Private Declare Sub CopyMemoryWrite Lib "kernel32" Alias _
    "RtlMoveMemory" (ByVal Destination As Long, _
    source As Any, ByVal Length As Long)
Private Declare Sub CopyMemoryRead Lib "kernel32" Alias _
    "RtlMoveMemory" (Destination As Any, _
    ByVal source As Long, ByVal Length As Long)

Then the routine that does the work.

Public Sub Init()
    Dim ptrToDouble As Long
    Dim byteArray(7) As Byte
    Dim i As Integer
    
    byteArray(7) = &H7F 'maxdbl
    byteArray(6) = &HEF
    For i = 0 To 5
        byteArray(i) = &HFF
    Next
    ptrToDouble = VarPtr(MaxDouble)
    CopyMemoryWrite ptrToDouble, byteArray(0), 8

    byteArray(7) = &HFF 'mindbl
    byteArray(6) = &HEF
    For i = 0 To 5
        byteArray(i) = &HFF
    Next
    ptrToDouble = VarPtr(MinDouble)
    CopyMemoryWrite ptrToDouble, byteArray(0), 8
    
    byteArray(7) = &H7F '+inf
    byteArray(6) = &HF0
    For i = 0 To 5
        byteArray(i) = 0
    Next
    ptrToDouble = VarPtr(PosInfinity)
    CopyMemoryWrite ptrToDouble, byteArray(0), 8

    byteArray(7) = &HFF '-inf
    byteArray(6) = &HF0
    For i = 0 To 5
        byteArray(i) = 0
    Next
    ptrToDouble = VarPtr(NegInfinity)
    CopyMemoryWrite ptrToDouble, byteArray(0), 8

    byteArray(7) = &H7F 'quiet NAN
    byteArray(6) = &HF0
    For i = 0 To 5
        byteArray(i) = &HFF
    Next
    ptrToDouble = VarPtr(QuietNAN)
    CopyMemoryWrite ptrToDouble, byteArray(0), 8

End Sub

Finally, a slice out of the Complex Numbers project demonstrating the use of some of these declarations.

Public Function IsInfinity(c As Variant) As Boolean
    Dim bRes As Boolean
    bRes = False
    If TypeName(c) = "ComplexNumber" Then
        If c.Real = PosInfinity And c.Imag = PosInfinity Then bRes = True
        If c.Real = NegInfinity And c.Imag = NegInfinity Then bRes = True
    Else
        If c = PosInfinity Then bRes = True
        If c = NegInfinity Then bRes = True
    End If
    IsInfinity = bRes
End Function

5 comments:

Anonymous said...: Your quiet NaN seems to have the first 4 bits of the mantissa being 0000. That's a signalling NaN, isn't it. I think your byteArray(6) should be set to &FF.; 11:53 am
Rick Regan said...: You've defined mindbl as -maxdbl. Usually, DBL_MIN (as it would be called in C) is the smallest POSITIVE number -- actually, the smallest normalized positive number -- which is 2^-1022.

In 'raw' format, 2^-1022 is the floating-point number with a sign field of 0, an exponent field of 1, and a fraction field of 0 (see my post http://www.exploringbinary.com/what-powers-of-two-look-like-inside-a-computer/ for more info).; 10:06 pm
Doctor Binary said...: You've defined mindbl as -maxdbl. Usually, DBL_MIN (as it would be called in C) is the smallest POSITIVE number -- actually, the smallest normalized positive number -- which is 2^-1022.

In 'raw' format, 2^-1022 is the floating-point number with a sign field of 0, an exponent field of 1, and a fraction field of 0 (see my post http://www.exploringbinary.com/what-powers-of-two-look-like-inside-a-computer/ for more info).; 12:02 am
Unknown said...: There is a much simpler way to get +infinity and -infinity.
Dim fPosInf As Double, fNegInf as Double
on error resume next
fPosInf=1/0
fNegInf=-fPosInf

From StackOverflow:
http://stackoverflow.com/questions/885994/how-do-you-get-vb6-to-initialize-doubles-with-infinity-infinity-and-nan/886148#886148; 2:00 pm
Bruce M. Axtens said...: Ah, yes, that link ... ummm ... you see, err, 'boost' is me. I posted this article and then went to StackOverflow in search of a better way of doing things. And they provided me with more than one.; 3:53 pm