Re: Question about out parameters of unconstrained array type.

[Phil Thornley <phil.jpthornley@gmail.com>]

In article <4e2c786f-8485-436f-b9c5-c8c19cf00805
@q12g2000yqg.googlegroups.com>, mark.lorenzen@gmail.com says...
> 
> On 25 Feb., 23:44, Phil Thornley <phil.jpthorn...@gmail.com> wrote:
> > > Would SPARK accept
> >
> > > ￜ ￜfor I in S'Range loop
> >
> > > ?
> >
> > No - the subtype of the loop index is still regarded as anonymous.
> 
> Phil
> 
> Do you know the rationale for this? Why can't the examiner just
> generate hypotheses like
> 
> loop__1__i <= s__index__subtype__1__last
> loop__1__i >= s__index__subtype__1__first
> 
> and then generate appropriate hypotheses regarding
> s__index__subtype__1__last and s__index__subtype__1__first?

Mark,

I don't know of any specific reasoning, but the rule that an explicit 
subtype mark must be given is a fundamental rule in the grammar, so I 
guess there would have to be a very substantial gain from modifying it 
before such a change was considered.

Cheers,

Phil