A curiosity about decimal fixed point types...

A curiosity about decimal fixed point types...

[mockturtle]

Dear all,
just a silly curiosity about decimal fixed point types, e.g.,

 type Euro is delta 0.01 digits 20;  
 -- Would be 20 enough? :-)

Are they currently used? Where? Since most of the examples are about money (mine and RM's) included, I guess that they were thought for financial applications, but I am unsure if nowadays financial software uses this type of types or just floating point.

Best regards

Riccardo

Re: A curiosity about decimal fixed point types...

[Mike H]

In message <7d4f2e0c-0aed-4ba3-aa5a-0c2bc0c2c5a0@googlegroups.com>, 
mockturtle <framefritti@gmail.com> writes
>Dear all,
>just a silly curiosity about decimal fixed point types, e.g.,
>
> type Euro is delta 0.01 digits 20;
> -- Would be 20 enough? :-)
>
>Are they currently used? Where? Since most of the examples are about 
>money (mine and RM's) included, I guess that they were thought for 
>financial applications, but I am unsure if nowadays financial software 
>uses this type of types or just floating point.
Having worked in financial areas before moving into Ada, I would suggest 
that accumulations of rounding errors are a potential mine field for the 
honest and a potential gold mine for the dishonest! In consequence 
rigorous algorithm specification is far more important than most people 
would expect.

Perhaps the ideal is to work in integers (exact figures of the lowest 
denomination of the coinage - pence, cents or w.h.y). But of course, too 
often this is not possible (e.g. currency conversions and discount 
rates) and that is when the problems really start.
-- 
Mike
"Homo sum, humani nihil a me alienum puto" =
"I am human, (so) I (should) judge nothing of humanity to be strange."
Publius Terentius (195/185–159 BC)

Re: A curiosity about decimal fixed point types...

[Oliver Kleinke]

Am Mon, 11 Nov 2013 07:04:12 -0800 (PST)
schrieb mockturtle <framefritti@gmail.com>:

> Dear all,
> just a silly curiosity about decimal fixed point types, e.g.,
> 
>  type Euro is delta 0.01 digits 20;  
>  -- Would be 20 enough? :-)
> 
> Are they currently used? Where? Since most of the examples are about
> money (mine and RM's) included, I guess that they were thought for
> financial applications, but I am unsure if nowadays financial
> software uses this type of types or just floating point.
> 
> Best regards
> 
> Riccardo
> 

Hi Riccardo,

The type definition that you provided is a good example for using fixed
point types for accurate representation. With a fixed-point type you
can avoid the occasional precision problems that floating-point types
suffer from. (For example if I type "0.1 + 0.2" in my
Python-interpreter it yields 0.30000000000000004.)
This can be very important if you sum up/integrate a range of
measurement data obtained from a sensor -- for example a gyro.
Also, on some machines fixed-point arithmetic 'might' be faster than
floating-point arithmetic -- e.g. machines without FPU -- but I guess
that also very much depends on the operations and the compiler, &c.

Re: A curiosity about decimal fixed point types...

[AdaMagica]

Look at
http://en.wikibooks.org/wiki/Ada_Programming/Types/delta
for a description of the differences between ordinary and decimal fixed point arithmetics.

Decimal has precise rules like Cobol (needed for currency), whereas ordinary is less precise.

Re: A curiosity about decimal fixed point types...

[Jacob Sparre Andersen]

Riccardo <framefritti@gmail.com> wrote:

>  type Euro is delta 0.01 digits 20;  
>  -- Would be 20 enough? :-)

Depends on your budget and your cash-flow.  For mine it is plenty.

> Are they currently used?

Yes.

> Where?

In 182 out of 119_391 Ada source files (including duplicates) on my laptop.

Reduced to unique type declarations, there is:

   type Decim is delta 0.1 digits 5;
   type Duration is delta 0.001 digits 9 range -24.0 * 3600.0 .. 48.0 * 3600.0;
   type Duration is delta 1.0 digits 9 range -24.0 * 3600.0 .. 48.0 * 3600.0;
   type Kroner is delta 0.01 digits 10;
   type Megabucks is delta 0.01 digits 15;
   type My_Float is delta 0.01 digits 10;
   type Percent is delta 0.01 digits 5 range 0.0 .. 100.0;
   type Push_Data_Type is delta 0.01 digits 7;
   type Real_Val_To_Print is delta 0.01 digits 8;

> Since most of the examples are about money (mine and RM's) included, I
> guess that they were thought for financial applications, but I am
> unsure if nowadays financial software uses this type of types or just
> floating point.

This indicates that they are used for a few more cases than just money
(2 of 9 are money types), but it appears that they aren't all that
common.

Greetings,

Jacob
-- 
"It's not a question of whose habitat it is,
 it's a question of how fast you hit it."

Re: A curiosity about decimal fixed point types...

[Marius Amado-Alves]

>    type Real_Val_To_Print is delta 0.01 digits 8;

I too use it for printing:-)

Re: A curiosity about decimal fixed point types...

[gautier_niouzes]

> >    type Real_Val_To_Print is delta 0.01 digits 8;
> I too use it for printing:-)

Since it is useful for money and for printing, then it should be perfect for money printing - a big business these days!...
_________________________ 
Gautier's Ada programming 
http://gautiersblog.blogspot.com/search/label/Ada 
NB: follow the above link for a valid e-mail address