[Python-Dev] Decimal data type issues

[Andrew P. Lentvorski, Jr.]

On Tue, 13 Apr 2004, Batista, Facundo wrote:

> The issue is that this limit is artificial: As long it's a long, you should
> be able to make it as big as your memory let you.

I agree with Tim that underflow/overflow are useful flags of something
going wrong.  Very few people would connect Decimal overflow with "Gee,
why is Python allocating all of my system's memory?"  I would like to see
a soft default limit that can be modified.

As for the value of the default limits, I would actually choose both a
default precision and exponent limit which would be within the range
represented by a double precision floating point number.

This helps in two cases:

1) People will try to interconvert FP and Decimal at various points.
Having one with a significantly different range will certainly give some
surprises. (example: Decimal would get converted to double precision FP
before being used in OpenGL or Direct3D to draw a pie chart, graph, etc.)

2) Having a default range somewhere inside double precision means that
some mythical "efficient" implementation could actually use double
precision FP operations to get close and then use a cleanup step to check
the final digit.  (Very useful for exponentials)

-a