[texhax] why not floating-point?

[Barbara Beeton]

    What about adding floating-point arithmetic support to TeX?

perhaps it could be considered for an extension to a
tex successor.  but it will never be added to tex
itself.  when tex was implemented on multiple platforms,
it was demonstrated that the floating-point arithmentic
was seriously not the same on all platforms, hence the
decision to use only a very fine-grained fixed-point
mechanism.  perhaps the situation has changed now, and
floating point is (more) uniform.  but that would need
to be demonstrated, to ensure platform independence.

    Motivation:  I understand that the main obstacle to implementing truly
    powerful graphic routines directly via TeX is that TeX cannot handle
    floating-point arithmetics.  While various external tricks are
    wonderful, they seem a bit unstable (eg the beauty of PostScripting via
    packages like eepic, that suddenly becomes unavailable when switching
    to pdfTeX).

a worthy goal, but it doesn't outweigh the original
motivation, which strongly affects the positioning of
letters in a line of type or symbols in a built-up
equation.

in any event, tex itself is frozen, so this is a
topic for a successor.
							-- bb