home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!snorkelwacker.mit.edu!ai-lab!life.ai.mit.edu!tmb
- From: tmb@arolla.idiap.ch (Thomas M. Breuel)
- Newsgroups: comp.lang.c++
- Subject: Re: Power Operator vs NANs
- Message-ID: <TMB.92Jul21215436@arolla.idiap.ch>
- Date: 22 Jul 92 01:54:36 GMT
- References: <819@nazgul.UUCP> <1992Jul15.123527.3771@bnr.co.uk> <850@nazgul.UUCP>
- <14hm92INNrt0@agate.berkeley.edu>
- Sender: news@ai.mit.edu
- Reply-To: tmb@idiap.ch
- Organization: IDIAP (Institut Dalle Molle d'Intelligence Artificielle
- Perceptive)
- Lines: 21
- In-reply-to: jbuck@forney.berkeley.edu's message of 21 Jul 92 18:50:10 GMT
-
- In article <14hm92INNrt0@agate.berkeley.edu> jbuck@forney.berkeley.edu (Joe Buck) writes:
-
- One could still implement the comparison operators that you
- describe [i.e., the new NCEG comparison operators] that consider
- the possibility of unordered values (NaNs) and have the same syntax
- work on machines that don't have NaNs. It would just be that the
- test for two values to be unordered would never come out true on
- such machines.
-
- I remain unconvinced (even after having read the NCEG document),
- however, that an algorithm that needs to rely on the new NCEG
- comparison operators will do anything reasonable on architectures that
- don't support NaN's or Inf's.
-
- Frankly, I'd much rather have an error flagged at compile time
- ("isnan() not defined") than wonder why some numerical algorithm out
- of some library goes into an infinite loop once in a blue moon when I
- compile it on my non-IEEE machine.
-
- Thomas.
-
-