standards/112180: tgamma on negative integers returns Inf instead of NaN

[Bruce Evans]

The following reply was made to PR standards/112180; it has been noted by GNATS.

From: Bruce Evans <bde at zeta.org.au>
To: =?X-UNKNOWN?Q?Vincent_Lef=E8vre?= <vincent at vinc17.org>
Cc: freebsd-gnats-submit at FreeBSD.org, freebsd-standards at FreeBSD.org
Subject: Re: standards/112180: tgamma on negative integers returns Inf instead
 of NaN
Date: Sun, 29 Apr 2007 10:14:55 +1000 (EST)

   This message is in MIME format.  The first part should be readable text,
   while the remaining parts are likely unreadable without MIME-aware tools.
 
 --0-200284211-1177805695=:17746
 Content-Type: TEXT/PLAIN; charset=X-UNKNOWN; format=flowed
 Content-Transfer-Encoding: QUOTED-PRINTABLE
 
 On Fri, 27 Apr 2007, Vincent Lef=E8vre wrote:
 
 > Though POSIX allows an implementation-defined value to be returned, this =
 is mainly in the case where NaN is not supported, and NaN is the best choic=
 e and is what tgamma returns on other platforms (Linux, HP-UX, Mac OS X).
 
 Also, Annex F of C99 (about the IEEE-754/IEC-mumble binding) requires a NaN
 plus the "invalid" exception.
 
 Bruce
 --0-200284211-1177805695=:17746--