[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: Possible Wording Contradiction in Signaling Comparisons (section 5.12)
You are correct, section 7 has not been updated here to reflect
what's in the current section 5.
I can make the change editorially.
The change is:
Comparison by way of predicates listed in Clause 5.12,
Table 4b: Required signaling predicates and negations,
when the operands are unordered.
I noticed what an apparent wording contradiction regarding signaling
comparisons, in section 5.12 of the June 26th draft. Namely, these
sentences disagree as to whether ==, != have signaling comparison
operations.
"... an invalid operation exception (7.1) shall be signaled when, as
indicated inTable 4, last column, unordered operands are compared
using one of the predicates involving < or > but not ? (Here the
symbol ? signifies unordered).
...
Table 4b lists the predicates that cause an invalid operation
exception
when the relation is unordered. That invalid exception defends
against
unexpected quiet NaNs arising in programs written using the six
standard
predicates (==, !=, <, <=, >, >=) without considering the
possibility
of a quiet NaN operand."
Perhaps I'm missing something here, but do we want signaling [==, !=]
functions or not? (As a point of reference, the original 754 draft does
not signal invalid exception for these two).
Also, the phrase "inTable 4, last column" is unclear to me. There are
three Table4's (4a, 4b, 4c) in the 754R draft, which specify different
operations. Is this reference to Table4 of the original 754 standard,
or the series of tables in the 754R draft?
Please advise on whether I'm missing something here, or if it seems
worthwhile to clarify this wording and the associated tables.