Tels <[EMAIL PROTECTED]> writes: >> >> > Seriously, how you you represent NaN as a FP? +-Inf is easier... >> >> sin(inf) seems to be the traditional way. I haven't found a way to >> get a negative nan though. > >Ah, damn. I shouldnt have asked - you are telling me there are negative >NanS? Ouch.. I shoulda have sticked to positive integers only...*sigh*
-ve NaNs are usually easy to generate - just negate a +ve one (most IEEE machines negate FP by togging sign bit without looking closely at rest of it). That said there is no semantic significance to a -ve NaN. There are however two classes of NaN - signaling and non-signaling. IIRC ops on a signaling NaN raise an exception while non-signaling NaNs just force result of any op to be non-signaling NaN (kind of like perl's tainting). > >Best wishes, > >Tels > >- -- > Signed on Thu Jan 6 21:22:10 2005 with key 0x93B84C15. > Visit my photo gallery at http://bloodgate.com/photos/ > PGP key on http://bloodgate.com/tels.asc or per email. > > http://www.informatik.hu-berlin.de/~hstamm/beweis.html - NÃtzliche > Beweismethoden fÃr jeden Tag > >-----BEGIN PGP SIGNATURE----- >Version: GnuPG v1.2.4 (GNU/Linux) > >iQEVAwUBQd2ePHcLPEOTuEwVAQEdvwf+L18DdJ2J+c6DOxBjrTrWBygdnPAqAzKH >TvUcdnwWnglXrNdexffDG5c40PV9x5tZppq7x2jYIMEk89jswukRjGCuE8jsqWSj >Uj6xaqYc7z7TXXzqbmrwkTVIpR5EDDf4vbdlc+FQKSJoJBjyn3TarfB9kn98mfMf >awUO8h2LMA+1oAnTnfvlKx1wKL8jX1dvWeO4ICeXEPTQ9fo8Odxh3UpkNhI8hRT6 >KduEPOrr9vcy0dBNUOsuTJ0VHc97oOwkNF1CmQaJDcomyrO56wSN86W4ZERhhX+f >oehlyLlbfnivKm495LygEnuHmHfnonQlyAhvqG1dp9djmBQRusCUDg== >=eqc/ >-----END PGP SIGNATURE-----
