Re: [xslt] floor(), ceiling(), and round() functions and negative values

[Bjorn Reese <breese mail1 stofanet dk>]

Daniel Veillard wrote:

>   libxslt doesn't make a difference between -0 and 0, right this may
> be considered a bug ... The associated questions are:
>     - does anybody care ?

Not me.

However, the XPath specification says (section 4.4) that "If the
argument is negative zero, then negative zero is returned."


>     - does anybody knows how to code the detection in a way
>       portable across the set of architectures libxslt support ?

trionan already contains that capability (trio_fpclassify_and_signbit()
and trio_nzero())