Re: [xml] Getting crazy with float parsing

[Bjorn Reese <breese mail1 stofanet dk>]

Peter Jacobi wrote:

> Whereas the standard runtimes (libc and msc) give
> the uninitiated human observers some surprise, they
> observe IEEE 754 and output a result differing only
> in the least significant digit from the true result.
>
> This can clearly be seen in 2.0/3.0 example, where the
> trio error is about 9 times higher than the others.

You omitted to calculate the error of the two other cases.

The higher error for some numbers is a due of the need for more binary
digits to round numbers more appropriately for non-binary radixes
(specifically decimal numbers).

Before making any decision on this issue, we should clarify the purpose
of XPath and XSL floating-point numbers. Is it to provide high-accuracy
calculations, or to present decimal-point number to users?

In the former case we should probably consider switching to 'long
double'. In the latter case common knowledge tells us that 0.667 is
a more correct representation of 2/3 than 0.6663, even if it is less
accurate.