[Cfp-interest 2510] JT-091 through JT-097 comments on N3047

[Jim Thomas]

Here are some more draft comments on C23 review draft N3047, from reviewing 7.27 and 5.2.4.2.2.

- Jim Thomas


JT-091 7.27#4 There is a spurious line break after “modf”.

JT-092 7.27#15 After the table there is an inadvertent list of macro names “acospi asinpi … dsqrt”.

JT-093 5.2.4.2.2#13 Since “range” is used in multiple ways, and “range of representable values” has the specific meaning given in this paragraph, it would be better to replace the two instances of “the range of that type” with “the range or representable values of that type”.

JT-094 5.2.4.2.2#14 Change “… and of the library functions …” to “and of most of the library functions …”. Some of the library functions have fully specified result accuracy, e.g. fabs and creal. 

JT-095 5.2.4.2.2#14 This paragraph is about operations, not type characteristics. so doesn't seem to belong in 5.2.4.2.2. However, I don’t see a good alternative that doesn’t entail repeating the same words in several places.

JT-096 5.2.4.2.2#18 Problem 1: The wording is awkward or worse. The first part indicates two alternatives where the second alternative “perhaps" has an extra feature. But there are three values. 
Problem 2: A parenthetical remark says “(this does not imply conformance to Annex F)”. But what does matches IEC  60559 operations mean? Can the binding be something different from what’s in Annex F? It seems to offer a way to claim IEC 60559 support without saying what it is.
To address both problems, delete “(and  perhaps, operations)” and also delete  "2 type matches an IEC 60559 format and operations”.

JT-097 5.2.4.2.2#26 Footnote 28 says "If the presence or absence of subnormal numbers is indeterminable, then the value is intended to be a positive number no greater than the minimum normalized positive number for the type.” The presence or absence of subnormal numbers being indeterminable is one of the problematic conditions that the now obsolescent type_HAS_SUBNORM macros attempted to characterize. Also, the direction given by the footnote is not helpful since it is clearly implied by the definition of the type_TRUE_MIN macros. This footnote should be removed. Note that the footnotes that attempted to clarify the specification of the type_HAS_SUBNORM macros have already been removed.