Re: arithmetic issues Thomas Bushnell BSG 23 Oct 2005 00:47 UTC

Aubrey Jaffer <xxxxxx@alum.mit.edu> writes:

>  | From: Thomas Bushnell BSG <xxxxxx@becket.net>
>  | Date: Fri, 21 Oct 2005 18:22:45 -0700
>  |
>  | Aubrey Jaffer <xxxxxx@alum.mit.edu> writes:
>  |
>  | > I think that an implementation should be allowed to signal an
>  | > error under some conditions where an error object is encountered.
>  | > Mandating readable written representations for error objects
>  | > prevents an implementation from signaling such errors.
>  |
>  | I think this might be confused.  Surely the mandating of a
>  | representation would mean "if you print something (rather than
>  | signalling an error) you should print it such-and-such a way."
>
> That still prevents an implementation from displaying information
> about what type of NaN was returned.  Such information could be
> helpful to find the call which generated the NaN.

Huh?  How does it prevent such?  We *could* mandate a readable written
representation for NaNs without manding that printing a NaN should
produce that representation, since it would still be allowed to signal
an error.  (And then, once it is signalled, it could print *anything
it wants*.)

Moreover, nothing prevents the mandated written representation from
optionally including implementation defined contents, if that should
be useful.

Thomas