Hi Eike,

Am Mittwoch, 21. November 2007 17:27 schrieb Eike Rathke:
> Hi Andreas,
>

[...]

> Ah, yes, the dark chapter of day count algorithms.. in the OASIS
> OpenDocument Format Formula specification subcommittee we struggle with
> that as well, namely for the YEARFRAC() function and all related
> financial functions that are based on it.
>

Google produces some links, the most recent of which seems to be

http://jackwaugh.com/it/openformula-20070125.html#6.9.23.YEARFRAC|outline

which is not even on oasis-open.org. Since I didn't find anything more recent, 
I am referring to this source.

Some day count conventions differ only slightly -- and only in some particular 
cases. So I think it will not be enough to literally describe the 
conventions. It will not be enough to use identifiers such as act/act, 30/360 
etc., either. At least some of them do not seem to be defined by a normative 
source and different sources associate different conventions with these 
identifiers -- or vice versa.

But even if there are normative sources -- not everybody will always have easy 
access to them.

So I strongly advise, to use mathematical formulas in the specification to 
define the conventions. They will then also have a chance to make it into the 
online documentation of applications implementing the spec.

The description in 

http://www.swx.com/download/trading/products/bonds/accrued_interest_en.pdf

could serve as an example.

I guess OO, in particular calc will implement the OpenFormula specification. 
Do you have any noncommittal indication about when the specification might be 
finalized -- and be implemented by OO?

The source given above indicates, that OO 2.0.3 already has a yearfrac 
function.  I have OO 2.0.4 but it doesn't know yearfrac...? Well, I have a 
german version, maybe the function has a different name which I cannot find 
out.

(It might be a good idea to >not< translate function names... at least to 
always support the English names in any language whatsoever.)

- Andreas

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to