(j3.2006) (SC22WG5.4156) [ukfortran] Question about our design for associate names

[N.M. Maclaren]

On Feb 5 2010, Bill Long wrote:
>
>> Do we want to
>> 
>> (1) add the restrictions to associate names
>>     (a) by way of an interp against 2003,
>>     (b) in 2008 and announce an incompatible change, or
>>     (c) in 2013 and announce an incompatible change, or
>> (2) not add the restrictions?
>
>I would certainly vote for (2).  It has two major advantages:  It does 
>not change the current concept that associate constructs are just a 
>mechanism for creating aliases to simplify expressions, and, more 
>important, it would not suddenly invalidate existing codes.

Yes.  Keep it simple.  While I dislike the way that pointers permit
aliasing in constructs that would otherwise not allow it, I don't think
that any more special cases would be helpful.  I took a look at that
area, from the point of view of teaching it, and decided not to!  The
current rules are already too complicated to explain in simple terms.

If this were being tackled, I would prefer a more generic approach,
such as a new attribute (say PURE), that would apply to POINTER and
TARGET and remove almost all of the aliasing liberty (i.e. the rules
for ordinary variables would apply).  Obviously not for 2008, and I
haven't thought this through, so it might not fly.

Regards,
Nick.