		Release notes for FPLIB Release 2.0
		===================================

This release contains fixes to FPLIB release 1.0, and fixes to the
standard Sozobon release 1.0 floating support.

In the standard support:

- The original floating point add/subtract, multiply, and divide have
  been rewritten for better speed and accuracy.

- Bugs in the floating version of printf have been fixed.

The remaining information is of interest only if you have been using
FPLIB release 1.0.  The following changes have been made:

- MATH.H has been expanded to provide symbolic definitions of some
  common constants.

- A few minor bugs have been fixed:

  * Some routines could return -0.0; the sign has been corrected.

  * Some routines could return a number with a biased exponent of
    zero; they now return true 0.0.

  * Out of range results are now signaled by ERANGE, not EDOM.

  * frexp() returned an incorrect value on overflow, and ldexp()
    didn't handle overflow or underflow correctly.
