Use of expression function in protected type

Use of expression function in protected type

[Simon Wright]

Is it permissible to use an expression function as the completion of a
protected function? (GNAT thinks so).

   protected Button
   is
      function Current_Index return Interval_Index;
   private
      Index : Interval_Index := 0;
      procedure Handler;
      pragma Attach_Handler (Handler, Ada.Interrupts.Names.EXTI0_IRQ);
   end Button;

and then

   protected body Button is
      function Current_Index return Interval_Index is (Index); -- <<<<<<<<
      procedure Handler is
      begin
         HAL_GPIO_EXTI_IRQHandler (16#0001#);
         Index := Index + 1;
      end Handler;
   end Button;

ARM 6.1(30) distinguishes an expression_function_declaration from a
subprogram_declaration.

6.8(4) allows an expression_function_declaration to be a completion.

9.4(8) says a protected_operation_item can be, inter alia, a
subprogram_declaration or a subprogram body. (subprogram_declaration?
how can that be?)

Re: Use of expression function in protected type

[Egil H H]

On Tuesday, January 20, 2015 at 11:34:54 PM UTC+1, Simon Wright wrote:
> 9.4(8) says a protected_operation_item can be, inter alia, a
> subprogram_declaration or a subprogram body. (subprogram_declaration?
> how can that be?)

Just like a package can have subprograms declared
in the public part, private part or the body,
protected subprograms can be declared in the public 
part, private part or body of a protected type.
(For example, it's not uncommon for a barrier 
function to be declared in the body.)

And just like in a package body, forward 
declarations (or subprogram_declarations) are 
allowed for subprograms. 

However, an expression_function is allowed 
to complete a subprogram_declaration, but is not
itself a subprogram_declaration (ARM 6.1(30/3).
As far as I can tell, it's a basic_declaration,
which is allowed in package specifications and 
bodies, but not in protected_operation_items.

-- 
~egilhh

Re: Use of expression function in protected type

[Simon Wright]

Egil H H <ehh.public@gmail.com> writes:

> On Tuesday, January 20, 2015 at 11:34:54 PM UTC+1, Simon Wright wrote:
>> 9.4(8) says a protected_operation_item can be, inter alia, a
>> subprogram_declaration or a subprogram body. (subprogram_declaration?
>> how can that be?)
>
> Just like a package can have subprograms declared
> in the public part, private part or the body,
> protected subprograms can be declared in the public 
> part, private part or body of a protected type.
> (For example, it's not uncommon for a barrier 
> function to be declared in the body.)

One lives and learns! I've never had occasion to write a barrier
_function_; my most complex barrier was
   when A
     or else not B
     or else (C and then not (D and then E))

> However, an expression_function is allowed 
> to complete a subprogram_declaration, but is not
> itself a subprogram_declaration (ARM 6.1(30/3).
> As far as I can tell, it's a basic_declaration,
> which is allowed in package specifications and 
> bodies, but not in protected_operation_items.

That was my reading, but I wondered whether it was deliberate, an
oversight, or I'd missed something.

The context was Emacs ada-mode, whose indentation engine is built on a
parser, which follows the ARM.

Re: Use of expression function in protected type

[Randy Brukardt]

"Simon Wright" <simon@pushface.org> wrote in message 
news:ly8ugwo5pm.fsf@pushface.org...
...
>> However, an expression_function is allowed
>> to complete a subprogram_declaration, but is not
>> itself a subprogram_declaration (ARM 6.1(30/3).
>> As far as I can tell, it's a basic_declaration,
>> which is allowed in package specifications and
>> bodies, but not in protected_operation_items.
>
> That was my reading, but I wondered whether it was deliberate, an
> oversight, or I'd missed something.

Certainly not deliberate, I'm pretty sure no one ever considered it. We will 
now (I've forwarded a version of your message to Ada-Comment).

                     Randy.

Re: Use of expression function in protected type

[Randy Brukardt]

I said:

> Certainly not deliberate, I'm pretty sure no one ever considered it. We 
> will now (I've forwarded a version of your message to Ada-Comment).

Early returns are that it is an oversight. It'll be on the agenda for next 
week's ARG phone call, and quite possibly it will get included in the 
upcoming Corrigendum. If so, it probably will be close to the fastest 
official Ada fix ever...

                                Randy. 

Re: Use of expression function in protected type

[Simon Wright]

"Randy Brukardt" <randy@rrsoftware.com> writes:

> I said:
>
>> Certainly not deliberate, I'm pretty sure no one ever considered
>> it. We will now (I've forwarded a version of your message to
>> Ada-Comment).
>
> Early returns are that it is an oversight. It'll be on the agenda for
> next week's ARG phone call, and quite possibly it will get included in
> the upcoming Corrigendum. If so, it probably will be close to the
> fastest official Ada fix ever...

Wow!