(j3.2006) pointer arguments to PURE procedures

[Tobias Burnus]

Hi Malcolm,

On 09/26/2012 10:08 AM, Malcolm Cohen wrote:
> Maybe. I've not seen "all" respond...

I have to admit that Toni All hasn't responded - not that an artist's 
opinion would carry specific weight; but  what I really meant was that 
"all who have written about the issue" seem to have come to the same 
conclusion.

> In this situation there is no alternative to doing the grunt work that 
> is going through the draft revisions of F2008 and finding which one 
> has the new (bad) wording - you will want to use binary chop on that 
> task. Then looking at the editor's report and the papers that went 
> into that draft, any paper that makes a technical change must describe 
> what technical change is being attempted or at least reference back to 
> an earlier paper which formally describes the technical change.

Very good point. (Sometimes one misses the obvious!) I really tend to 
forget which information is on the J3 server and how j3/wg5 
standardization works, which makes such information publicly available. 
(Contrary to some other specs where one only gets dumped the final spec.)

The difference is between 08-007r2 and 09-007, affecting C1280. The 
change was proposed at http://j3-fortran.org/doc/year/08/08-181.txt and 
the editor's report is 09-102.

 From the wording of the latter, it seems that someone had forgotten 
about intents on pointers:

   - [308:12.7 C1280] Instead,
       - deleted "is a dummy argument of a pure function,"
       - changed "INTENT (IN) of a pure subroutine"
         to "the INTENT (IN) attribute".
     Unless I am very much mistaken, this has the same effect and is much
     shorter and simpler.  That is, instead of the requested
       "is a dummy argument of a pure function that does not have the VALUE
        attribute, is a dummy argument with INTENT (IN) of a pure
        subroutine,"
     it now simply reads
       "is a dummy argument with the INTENT (IN) attribute,".

Thus, it seems to be unintended. I will write an IR.

Tobias