To: J3 Members J3/17-117
From: Van Snyder
Subject: Comments concerning atomic subroutines
References: 17-007
Date: 2017 February 01
1. Introduction
===============
Several of the atomic subroutines in 16.9.20 - 16.9.30 include a
requirement of the form "The values of VALUE and ATOM + VALUE shall be
representable in ATOMIC_INT_KIND."
These are nonsense because ATOMIC_INT_KIND is a named constant, if the
phrase actually refers to the ATOMIC_INT_KIND defined in the intrinsic
module ISO_FORTRAN_ENV. Further, if one interprets the requirement to
mean "The values of VALUE and ATOM + VALUE shall be representable as
integers of kind ATOMIC_INT_KIND" there's no requirement that the value
of ATOMIC_INT_KIND be the one from the intrinsic module ISO_FORTRAN_ENV.
2. Edits
========
[360:19 16.9.20p3] [360:34 16.9.21p3] [362:5 16.9.24p3]
[362:22 16.9.25p3] [362:41 16.9.26p3] [363:15 16.9.27p3]
[363:33 16.9.28p3] [364:21 16.9.30p3]
Replace "in ATOMIC_INT_KIND" with "as integers of the kind specified by
the named constant ATOMIC_INT_KIND from the intrinsic module
ISO_FORTRAN_ENV" eight times.