access types and head pains

access types and head pains

[chris.danx]

hi,
    I have the following situation.  I'm writing a driver that can be run on my
system.  I want to test it by writing virtually to a file.


The problem is i had the following code


package X is

    type A is private
    type A_ac is access all A;
...

    procedure Do_S (av : in out A_ac; v : any_type );

    --  with body
    --    is begin
    --        av.v := v; etc

end X;

procedure test is
    at : aliased A;
begin
    Do_S (at'unchecked_access, 2);
end test;

i got an error that said "at actual must be a variable".  What's wrong?

The idea is that the driver will work with two different kinda things in the
same procedure.  "at" can be declared in program or be a chunk of memory of the
same size. Any hints?


Thanks,
Chris

Re: access types and head pains

[Jeffrey Carter]

"chris.danx" wrote:
> 
> package X is
> 
>     type A is private
>     type A_ac is access all A;
> ...
> 
>     procedure Do_S (av : in out A_ac; v : any_type );
> 
>     --  with body
>     --    is begin
>     --        av.v := v; etc
> 
> end X;
> 
> procedure test is
>     at : aliased A;
> begin
>     Do_S (at'unchecked_access, 2);
> end test;
> 
> i got an error that said "at actual must be a variable".  What's wrong?
> 
> The idea is that the driver will work with two different kinda things in the
> same procedure.  "at" can be declared in program or be a chunk of memory of the
> same size. Any hints?

I suspect your message read "av actual ...". The parameter av is mode in
out, which means Do_S may modify it. You can't change
at'unchecked_access, because then it wouldn't be at'unchecked_access any
more. Unless you have "av := ...;" in Do_S, why not just make it mode
in?

-- 
Jeff Carter
"I waggle my private parts at your aunties."
Monty Python & the Holy Grail

Re: access types and head pains

[chris.danx]

> I suspect your message read "av actual ...". The parameter av is mode in
> out, which means Do_S may modify it. You can't change
> at'unchecked_access, because then it wouldn't be at'unchecked_access any
> more. Unless you have "av := ...;" in Do_S, why not just make it mode
> in?

I'll try that.  I'm a little confused though.

(let's change the name "at" to al_t, i quickly wrote down the example with short
identifiers of which at happened to appear, i know it's a reserved word!  i was
tired)

a_ac    is a general access type

if av has mode "in" then how can i modify the fields of the record.  surely this
would be dissallowed or does it mean you can't modify it's existance/null status
but can modify it's data?


Chris

Re: access types and head pains

[Jeffrey Carter]

"chris.danx" wrote:
> 
> a_ac    is a general access type
> 
> if av has mode "in" then how can i modify the fields of the record.  surely this
> would be dissallowed or does it mean you can't modify it's existance/null status
> but can modify it's data?

Given

Av : in A_Ac

you can't modify Av; within the subprogram Av is a constant. You can,
however, modify the value designated by Av (Av.all); since A_Ac is not
"access constant", Av.all is not a constant.

-- 
Jeff Carter
"You tiny-brained wipers of other people's bottoms!"
Monty Python & the Holy Grail

Re: access types and head pains

[chris.danx]

> Given
>
> Av : in A_Ac
>
> you can't modify Av; within the subprogram Av is a constant. You can,
> however, modify the value designated by Av (Av.all); since A_Ac is not
> "access constant", Av.all is not a constant.

thanks for that explanation.  I understand now



Chris Campbell