IEEE 754 (or 854) floating point arithmetics with MathML

In a thread initiated by Tobias Burnus (March 14, 2001) an inconsistency in CN types is highlighted. Namely, the description of cn:

http://www.w3.org/TR/MathML2/chapter4.html#contm_cn

does not list float amongst the allowed types (just real).

but

the general description of the type attrib:

http://www.w3.org/TR/MathML2/chapter4.html#contm_typeattrib

does include float. David Carlisle promised to raise this question with the working group. Since then (march 2001), no change. Could anyone explain what is the current "state of art". Has FLOATING-POINT atribute been discarded? How can I encode IEEE 754 floating point arithmetic with MathML?

Best regards.

------------------------------------------------------------------------------------
Zdenek Hurak, Center for Applied Cybernetics, 
Czech Technical University, Prague, Czech Republic
z.hurak@c-a-k.cz
------------------------------------------------------------------------------------