"Lawrie Schonfelder" <[log in to unmask]> wrote:
...
> This would be a very easy restriction to relax. Given the struct%comp form
> there is a little more text to explain that struct(i,j)%comp(k,l) associates
> with array(k,l,i,j) but this is better than not allowing the association at
> all. WE cann't rectify the struct%comp ordering mistake but we can fix the
> banning of arrays of arrays error.
Given FORALL, is it really necessary to bother? FORALL allows
one to specify the association of the struct's subscripts with those
of other arrays in any order that the program requires.
--
J. Giles
|