(j3.2006) (SC22WG5.5358) [ukfortran] From a colleague

Malcolm Cohen malcolm
Wed Oct 29 20:55:36 EDT 2014

>Obviously, the following code fragments have equivalent problems with NaNs:
>    IF (x) 123,456,789
>    IF (x < 0.0) GOTO 123
>    IF (x > 0.0) GOTO 789
>    GOTO 456

Obviously, they are not equivalent and so do not have equivalent problems.

>I am getting increasing tired of the polemic in favour of IEEE 754's
>handling of NaNs in conditionals.

There is no "polemic" here, we are just conforming to International Standards. 
Which just happen to be implemented on the vast majority of relevant computing 
systems so you are stuck with that anyway.

We are not going to pick a fight with the IEEE-754 committee.

>While is it certainly possible to claim that (say) NAG Fortran's traps
>are not in breach of IEEE 754

I question whether this is a suitable forum for a user 
complaint/query/whatever-you-want-to-call-it about the NAG compiler.

>, one has to twist the tail of the IEEE
>standard in order to do so.

We provide a means of specifying the initial set of halting modes.  There is 
nothing nonconforming about that.  We even helpfully name the options 
"-ieee=full" and "-ieee=stop"!  Within the program, the user has full control 
(regardless of the option he selected).

Anyway, this is way off topic for WG5.  If you have a query about the NAG 
compiler, I recommend contacting NAG user support.

................................Malcolm Cohen, Nihon NAG, Tokyo. 

More information about the J3 mailing list