Re: Decimal versus binary arithmetic was Re: J4 - presentation/discussion on "Future of the COBOL Standard"

On Mon, 7 Apr 2008 22:29:07 -0700 (PDT), Richard <riplin@[EMAIL PROTECTED]
>
wrote:

>On Apr 8, 2:11 pm, Robert <n...@[EMAIL PROTECTED]
> wrote:

>> It's of no use when dealing with integers, because they have no
fractional part.
>
>But they do have fractional parts.
>
>If I have my stock  valuation total in cents and I have the quantity
>in units then when I divide to find the average cost per unit I have a
>fractional part to deal with. This may be rounded using various
>methods but accounting practice requires this be done in decimal, and
>binary rounding does not give accurate decimal rounding.

An average is an approximation. You're not going to book the number, so
accountants don't
care whether it's off by a cent. 

I doubt you can find a requirement for decimal arithmetic in IFRS (EU) or
FASB (US)
accounting standards. Generally, standards define results and avoid
telling you how to
achieve them. 

I am very familiar with EDGAR, the US Securities and Exchange Commission
database for
electronic filing of mutual fund holdings, and the Thomson CDA/Spectrum
Institutional
13(f) Common Stock Holdings and Transactions database. If you go to Yahoo
Finance or
similar, pull up a mutual fund and click on Holdings, the numbers you will
see came
through those databases and my software. You will not see price per unit,
but you will see
the marked to market value of each holding rounded to thousands. The
standard for
computing that number says in its entirety:

"You must round entries in Column 4 (Value) to the nearest thousand
dollars, and omit the
last three places."
http://www.sec.gov/info/edgar/edgarfm-vol2-v5.pdf