Re: fuzzy rounding, roundoff, roundup(), et cetera.

I coded a version of roundup(xx, pp)

It correctly handles all the tricky cases I know of, including:
  ru 0.24999999999999997 16
  ru 0.1250000001000003 16
  ru 0.49999999999999997 16
  ru 0.1 -15

It comes with a test harness, and an updated preliminary specification,
in my fpu_exceptions.c program.
See the "ru" and "ruspec" commands in particular.
Interested parties are invited to experiment with it.
If you find a case that doesn't seem to be handled correctly,
please let me know.

  https://www.av8n.com/cgit/FPU/tree/fpu_exceptions.c