To: J3 Members J3/16-138
From: Van Snyder
Subject: Result value paragraphs in Clause 14
Reference: 16-007
Date: 2016 January 25
1. Introduction
===============
In the "Result value" paragraph for some functions in Clause 14, the
result value is described to be provided for "all real X". The result
has nothing to do with the value of X. Rather, it is provided for all
kinds of X, or maybe only for all kinds of X for which
IEEE_SUPPORT_DATATYPE ( X ) has the value true.
2. Edits
========
At the following places
[448:2 14.11.43p5] [448:15 14.11.44p5] [448:31 14.11.45p5]
[449:22 14.11.47p5] [449:34-35 14.11.48p5] [450:10 14.11.49p5]
[450:25 14.11.50p5] [451:2 14.11.51p5] [451:16 14.11.52p5]
[451:31 14.11.53p5]
Replace "all real X" with "real X of all kinds" or "real X of all kinds
for which IEEE_SUPPORT_DATATYPE ( X ) has the value true" (whichever one
of those is correct).
3. Question without edits
=========================
[452:11 14.11.54p5] claims that "IEEE_SUPPORT_UNDERFLOW_CONTROL () has
the value true if the processor supports control of underflow mode for
all floating-point calculations". IEEE_SUPPORT_UNDERFLOW_CONTROL is
probably not provided for all floating-point calculations if the
processor has both IEEE arithmetic, and some other arithmetic. Do we
want the result to be true iff support for control of underflow mode is
provided for all floating-point calculations, or only for floating-point
calculations involving IEEE data objects of a kind such that
IEEE_SUPPORT_DATATYPE ( X ), where X is of that kind, is true?
Should this be "IEEE_SUPPORT_UNDERFLOW_CONTROL () has the value true if
the processor supports control of underflow mode for all floating point
calculations involving real data objects of a kind such that
IEEE_SUPPORT_DATATYPE ( X ), where X is of that kind, is true"?