Re: Motion 42: NO
Le mardi 12 février 2013 à 13:26 -0500, Michel Hack a écrit :
> Guillaume Melquiond wrote:
> > Note that I did not say that the definition of "ill" was incoherent.
> > I said that defining "ill" as meaning that a point-wise function has
> > an empty domain is useless from the point of view of the standard.
>
> Would it help if we said that _ill (and NaI) are to be used ONLY for
> tracking bad constructions?
Definitely. Though the words "bad constructions" might a bit
restrictive. If the user explicitly poisoned a memory location with a
NaI (e.g. to mark an uninitialized location), this is a valid use case
but perhaps not covered by those words.
That said, I don't think the current wording needs to be changed. The
first three lines of Section 8.8.3 are just fine in my opinion:
The "ill-formed" decoration ill propagates unconditionally
through arithmetic expressions. Namely, the decorated interval
result of a library arithmetic operation is ill-formed
(decorated with ill) if and only if one of its inputs is
ill-formed.
Nothing more, nothing less, is needed, as far as arithmetic expressions
are concerned.
Best regards,
Guillaume