<html><head><meta http-equiv="Content-Type" content="text/html charset=windows-1252"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;">On Jun 10, 2013, at 8:50 AM, Duncan Sands <<a href="mailto:baldrick@free.fr">baldrick@free.fr</a>> wrote:<br><div><div><br></div><blockquote type="cite"><div style="letter-spacing: normal; orphans: auto; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; widows: auto; word-spacing: 0px; -webkit-text-stroke-width: 0px;"><blockquote type="cite"><blockquote type="cite">If undef can't be achieved, it might still be possible to fold to a NaN<br></blockquote><blockquote type="cite">or some other special value.<br></blockquote><br>Folding to NaN sounds workable to me.<br></blockquote><br>So if Y is a well-chosen NaN (or will any NaN do?) then fadd X, Y will always<br>be equal to Y? Does the same go for fmul, fsub and fdiv?</div></blockquote><br></div><div dir="auto">If Y is any NaN, then the result of any IEEE-754 basic operation (add, sub, mul, div, sqrt, ... ) will be NaN (it will not “be equal to Y", because NaNs compare unequal to everything).</div><div dir="auto"><br></div><div dir="auto">- Steve</div></body></html>