Math errors in python

[david h]

the problem with BCD or other 'decimal' computations is that it either
doesn't have the dynamic range of binary floating point (~ +-10**310)
or if it has unlimited digits then there is a LOT of software cranking
to do the math, whereas binary floating point is in the hardware. If
you want the language to use binary floating point (fast) but do the
rounding for you, then fine, but then you will have problems using it
for any real numerical task because the issue of rounding is very
important to numerical analysis, and is done different ways in
different cases. Every time the language runtime rounds for you, it is
introducing errors to your computations that you may or may not want.
There is a large body of knowledge surrounding the use of IEEE 754
floating point representation and if the language diverges from that
then users who want to do numerical analysis won't use it.

another question: do you want the math package to round for you, or do
you want the IO package to do it only when you print? You will get
different results from each. I could imagine a language runtime could
have a switch that tells it to automatically round the results for
you, either in the math or the IO.