RE: [open-cobol-list] Cobol Precision

[Bill Klein]

Roger,
  I am not certain why you talk about "1000 decimal".  The actual '02
definition only requires 32 digit *plus* "exponent" precision.  As far as a
library to "do this" goes, you might want to check out:

 http://www2.hursley.ibm.com/decimal/#decNumber

("The decNumber package, an arbitrary-precision implementation of the
specifications in ANSI C, provides a reference implementation for the
arithmetic and encoding") 

> -----Original Message-----
> From: Roger While [mailto:address@hidden 
> Sent: Friday, May 27, 2005 4:08 AM
> To: Bill Klein
> Cc: address@hidden
> Subject: RE: [open-cobol-list] Cobol Precision
> 
> OK. Thanks for the info. I overlooked E.16.
> Although it is still not clear to me what this reference to 
> 1000 decimal
> places is. (wonders if the gmp library would be able to handle that)
> 
> Roger
> 
> At 19:14 26.05.2005 -0500, you wrote:
> >When "ARITHMETIC IS STANDARD" (new in the '02 Standard) is 
> in effect, then
> >there are rules (designed) to produce "portable and 
> predictable" results 
> >for
> >SOME - but not all - arithmetic statements and expressions.
> >
> >The "SIDI" (Standard Intermediate Data Item) is a 
> "conceptual" data item
> >used for intermediate results.  This is a >>>DECIMAL<<< type
> >"floating-point" (not Binary, not Hex, not anything like 
> what most people
> >are used to today) AND it isn't even a "data-type" that is 
> required to 
> >exist
> >in COBOL itself (from a user perspective).
> >
> >Truncation and precision is done in Decimal (not binary) 
> Floating Point.
> >
> >There is a discussion of all of this - including examples of 
> how this 
> >works
> >in
> >   "E.16 Standard arithmetic"
> >starting on page 763 of '02 Standard.
> >
> >    ****
> >
> >However, you might also want to look at a proposal included 
> in the current
> >revision work for the '08 Standard.  Check out:
> >
> >   http://www.cobolportal.com/j4/files/05-0047.doc
> >
> >IEEE is working on creating a "new" decimal floating point data type
> >(actually 3 different ones) and these MAY be easier to use 
> in COBOL (as an
> >intermediate data type) when/if available on various 
> hardware/software
> >platforms.
> >
> >    ****
> >
> >Does this actually answer you question?  If not, can you help me 
> >understand
> >it better?
> >
> >NOTE WELL:
> >   In both the '02 Standard and the draft revision, the 
> DEFAULT remains
> >"arithmetic is native" - so the implementor can do anything they want
> >(including truncating every operand and intermediate data 
> item to integers
> ><G> - although I don't think that would be very popular).
> >
> > > -----Original Message-----
> > > From: address@hidden
> > > [mailto:address@hidden On
> > > Behalf Of Roger While
> > > Sent: Thursday, May 26, 2005 5:37 PM
> > > To: Bill Klein
> > > Cc: address@hidden
> > > Subject: [open-cobol-list] Cobol Precision
> > >
> > > Clarification required :
> > > 8.8.1.3.1.1 Precision and allowable magnitude
> > > A standard intermediate data item has the unique value of
> > > zero or a value
> > > whose magnitude is in the range of
> > > 10-999 (1.000 000 000 000 000 000 000 000 000 000 0E-999)
> > > through
> > > 101000 - 10968 (9.999 999 999 999 999 999 999 999 999 999 9E+999)
> > > inclusive, with a precision of 32 decimal digits. A standard
> > > intermediate
> > > data item shall be truncated or rounded
> > > to fewer than 32 digits only as explicitly specified.
> > >
> > > What exactly does this mean ?
> 
> 
>