Re: Subprogram Renaming

[bobduff@world.std.com (Robert A Duff)]

In article <4kjkg6$4m2@krusty.irvine.com>,
Adam Beneschan <adam@irvine.com> wrote:
>    package q is
>       procedure j;
>    end q;
>
>    package body q is
>       procedure k;
>       procedure j renames k;
>    end q;
>
>Are j'address and k'address equal?  Is there anything in the RM that
>says one way or the other whether they must be equal?

No.  The RM does not require them to be equal, nor does it require them
to be unequal.

>How about j'access and k'access?

No, no requirement.

In fact, it's worse than that -- the RM does not even require that
P'Access = P'Access, for any procedure P.  See 4.5.2(13-13.a) and
3.10.2(39).

>I can't think of a legitimate reason to compare the two in actual
>code, in either case.  I'm just curious if the RM says anything about
>it, and if this would prevent the compiler from generating code for
>"j" that calls "k".

The compiler can generate code for j that calls k.  It can also generate
code for j that just *jumps* to the start of k (so the return
instruction in k returns directly to the caller of j).  Or, it can
generate no code for j, and make calls to j and k call to the same
address.

- Bob