Subj : Re: I need a routine to convert a real48 type into a double
To   : borland.public.cpp.borlandcpp
From : E. Kali
Date : Mon Aug 08 2005 10:33 am

Hi,
I've found the definition of the Real48 on the internet. So i wrote the 
following function.

/***************************************************************************
* Function     : ConvertReal48ToDouble
* Parameter(s) : szReal48Value - reald48 Value
* Return value : double - converted real 48 value
* Description  : Converts a real48 value to double.
*
*  NOTE:   REAL48 format
*           s      f            e
*  bits   47    48...8    7..0
*
*                     s       (e-129)
*          v = (-1)  *  2              * (1.f )
*
*         If e = 0, then v = 0.
*
****************************************************************************/
double COxymonfilereader::ConvertReal48ToDouble(unsigned char 
szReal48Value[6])
{
  double fResult(0);
  //strip the data out of real48 type
  int E( (int) szReal48Value[0]);
  int S( (int) szReal48Value[5] & 0x80);
  int F(((int) szReal48Value & 0x7FFFF0)>>4);
  if(E>0 && E<=255){
    float f(0);
    char szStrVal[10];
    _snprintf(szStrVal, 10, "1.%d",F);
    f = atof(szStrVal);
    fResult = pow(-1.0,S) * pow(2.0,(E - 129)) * f;
  }//if(E>0 and E<=255)
  return fResult;
}

It seems to work!
What do you think about this.

"Ed Mulroy" <dont_email_me@bitbuc.ket> wrote in message 
news:42f3e243$1@newsgroups.borland.com...
>> function Real48ToDouble(value: TReal48__): Double;
>
> What is 'value'?
>
> .  Ed
>
>> Tales Aguiar wrote in message
>> news:42f3aebb@newsgroups.borland.com...
>>I don't now if has a more easy way. I would create a function in .pas file
>> like...
>>
>> unit XXX;
>>
>> interface
>>
>> type
>>  TReal48__ = packed array[0..6] of byte;
>>
>> function Real48ToDouble(value: TReal48__): Double;
>>
>> implementation
>>
>> function Real48ToDouble(value: TReal48__): Double;
>> type
>>  PReal48 = ^Real48;
>> begin
>>  result := PReal48(@value)^;
>> end;
>>
>> end.
>>
>>
>> and use it...
>
>

.