(j3.2006) GB36

[John Reid]

Bill Long wrote:
> While we do not always follow it, there is a general bias toward putting
> Notes at the end of a subclause. Particularly a short one like this,
> where the gap between the normative text and the related Note is small.
> This would require slight rewording of the introductory line of the Note.

I prefer to leave it where it is. J3 can decide on the best position 
next week.

> Somewhat more problematic (and not related to the Note), the last part
> of the paragraph, ", and any further use of them is undefined behavior",
> seems both unnecessary and also wrong. If you later call the function
> again, and the pointer in question becomes well define again,

Not if the Fortran processor made copies and used different memory or 
used the same memory but used it for another purpose between the calls.

> it is
> certainly not the case that the use of that pointer is "undefined
> behavior" at that point. Is it not sufficient to just say that the
> pointer becomes undefined on return from the call (the previous words in
> the sentence).

I have changed the wording of the end of my note in an attempt to make 
it clear that it is pointers to the dummy argument that become 
undefined, not pointers to the actual argument.

Actually, I think "object" in line 3 of the para is ambiguous, but I 
don't have a good suggestion for disambiguation. At least my note should 
help.

Cheers,

John.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: gb36.txt
URL: <http://j3-fortran.org/pipermail/j3/attachments/20111005/eda0d970/attachment.txt>