Re: put of access type

[sjw <simon.j.wright@mac.com>]

On Aug 21, 8:29 am, "Dmitry A. Kazakov" <mail...@dmitry-kazakov.de>
wrote:
> On Thu, 20 Aug 2009 15:27:17 -0700 (PDT), sjw wrote:
> > On Aug 20, 7:58 pm, "Dmitry A. Kazakov" <mail...@dmitry-kazakov.de>
> > wrote:
>
> >> For example, the standard explicitly requires the array address to be one
> >> of its first element, and not of the array's dope.
>
> > Note for OP: that's 'Address, of course -- not 'Access
>
> No, it is exactly X'Address. See RM 13.3 (14). That adds headaches when you
> are going to implement a tricky storage pool.

?

What I meant was what 13.3(14) says. Sorry if I was unclear.

Martin suggested using the debugger; I've just been trying exactly
that, and it's quite confusing because the debugger knows a lot about
the types involved. GNAT warns about conversion between pointer-to-
unconstrained and system.address because they are of different sizes
(64 & 32), but it's hard to see where the extra bytes live.