other comments

[Vincent Lefevre <vincent@xxxxxxxxxx>]

In addition to the missing functions, here are the main comments
I submitted:

* Editorial / page 50 / subclause 9.2.1 / line 14

The second paragraph of 9.2.1 mixes up function names.

Replace

  For the functions exp, exp2, and expm1

by

  For the functions exp, exp2, and exp10

* Technical / page 50 / subclause 9.2.1 / line 20

As a square root is always non-negative, rSqrt(-0) should be +inf.
The fact that sqrt(-0) is -0 is not a reason to have rSqrt(-0) = -inf.
Indeed, though its result has a negative sign, sqrt(-0) is non-negative
(as -0 = +0). Also, if the user wants to compose functions in his
software, it is up to him to make sure that direct composition makes
sense; but when a function is defined in a standard, it should preserve
basic mathematical properties (whenever possible), in particular when
the result is exact (no rounding).

Replace

  rSqrt(+/-0) is +/-inf

by

  rSqrt(+/-0) is +inf

* Technical / page 51 / subclause 9.2.1 / line 42

"This implies that sinPi(-x) and -sinPi(x) are the same number
(or both NaN) for all x." is not true for roundTowardPositive and
roundTowardNegative.

Add "in symmetric rounding directions".

* Technical / page 50 / subclause 9.2.1

Nothing is said about the sign of 0 for acos(1) and acosh(1).

Add somewhere in this section: "acos(1) is +0" and "acosh(1) is +0".

* Editorial / page 57 / subclause 10.4 / line 32

The standard says: "changing the payload or sign bit of a quiet NaN",
but there are operations for which the sign bit of a NaN is significant
(e.g., copy). Also this is contrary to the first paragraph of 6.2.3
("An operation which propagates a NaN operand to its result and has a
single NaN as an input should produce a NaN with the payload of the
input NaN if representable in the destination format."), though this
one is only a "should".

Remove this transformation.

-- 
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.org/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)