<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content="text/html; charset=us-ascii" http-equiv=Content-Type>
<META name=GENERATOR content="MSHTML 11.00.9600.19326"></HEAD>
<BODY style="WORD-WRAP: break-word; -webkit-nbsp-mode: space">
<DIV dir=ltr align=left><BR> </DIV>
<BLOCKQUOTE
style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0000ff 2px solid; MARGIN-RIGHT: 0px"
dir=ltr>
<DIV></DIV>
<DIV
style='FONT-FAMILY: "Helvetica Neue"; MARGIN: 0px; LINE-HEIGHT: normal; font-stretch: normal'>Here’s
an issue I noticed when reviewing the C2X drafts:</DIV>
<DIV
style='FONT-FAMILY: "Helvetica Neue"; MARGIN: 0px; LINE-HEIGHT: normal; font-stretch: normal'><BR></DIV>
<DIV
style='FONT-FAMILY: "Helvetica Neue"; MARGIN: 0px; LINE-HEIGHT: normal; font-stretch: normal'>Previously,
floating constants were always evaluated at translation time, using default
rounding. Now, with TS 18661-1 and 2, constant rounding modes set by the
FENV_ROUND and FENV_DEC_ROUND pragmas affect floating constants within the
scope of the pragma. A C floating constant does not include a sign, which C
regards as a unary operator. Thus, rounding of a negated constant (e.g., -0.1)
is done before negation, which for rounding up or down is undesirable, and is
unlike the treatment of the corresponding string (“-0.1”) by the strtod family
as required by F.5. </DIV>
<DIV
style='FONT-FAMILY: "Helvetica Neue"; MARGIN: 0px; LINE-HEIGHT: normal; font-stretch: normal'><BR></DIV></BLOCKQUOTE>
<DIV
style='FONT-FAMILY: "Helvetica Neue"; MARGIN: 0px; LINE-HEIGHT: normal; font-stretch: normal'
dir=ltr><SPAN class=510283606-15052019></SPAN><FONT face=Arial><FONT
color=#0000ff><FONT size=2>I<SPAN class=510283606-15052019> suppose one
could argue that this gives the programmer the choice. (Or they
could use decimal arithmetic, of course
:-).)<BR><BR>Mike</SPAN></FONT></FONT></FONT><BR></DIV></BODY></HTML>