Re: Proposal for 2.0: New monetary data types

[Chris Travers <..hidden..>]

On Sat, Apr 3, 2010 at 5:14 AM, Alejandro Imass <..hidden..> wrote:
> On Fri, Apr 2, 2010 at 2:06 PM, Chris Travers <..hidden..> wrote:

> Awesome ideas. I have thought of taking this a step further and
> instead of having a base currency in a particular coin, a
> multi-currency system could have a high precision "base value unit".
> It's much like having a base coin, but in generic value units that
> have constant value. That is, your base VU will not fluctuate due to
> exchange rates and such. I know this dwelves into some complex
> economic theory but since you are evaluating ideas for multi-currency
> there is my .02.

I am not entirely sure how this would work in practice.

Also there are some minor changes to the above proposal that have
occurred since I initially sent it:

1)  The value component would be the value of the monetary element, so
monetary('1000', 'USD', '2') would be 1000 USD in $2 bills instead of
2000 USD.  This makes the math a lot easier.  It also allows for NULL
denominations where the money is not in a specific denomination (i.e.
a collection of cash in various denominations, a check, or a credit
card purchase).  The tradeoff is that if you end up with something
like coins worth 1/12 of a base currency rate, one could no longer
keep the precision exact.

2)  Two different currency exchange functions would be added:  The
first (part of the core module) would convert from one currency to
another at a specified rate.  The second (part of the business logic
module) would convert one currency to another as of a stored rate on a
specified date.

Also this will be broken off and implemented in C as a Pg-foundry
project and the goal would be to keep this applicable to other
software projects as well (or rather I am collaborating with someone
to do this).

Best Wishes,
Chris Travers