Re: Initialization and Finalization of limited object "returned" by a function

["Hibou57 (Yannick Duchêne)" <yannick_duchene@yahoo.fr>]

On 11 fév, 12:00, Ludovic Brenta <ludo...@ludovic-brenta.org> wrote:
> Yannick Duchêne wrote on comp.lang.ada:
>
> >    Tested_Interface : Dispatch.Instance_Class :=
> >      (Dispatch.New_Instance);
>
> Here you are initializing with an aggregate that contains the result
> of the function call. I suspect this triggers a compiler bug. What
> happens if you remove the parentheses?
As Jean-Pierre explained, this is not an aggregate, its a
parenthesized expression.

Let me explain why such a simple expression is parenthesized here :
this is just a matter of text layout. I follow one part of the
convention which is in use in the world of GNAT/GPS, which applies an
outdent of one character whenever a language construct which is able
to start with a parenthesis span multiple lines. This make sens for a
complex expression or an enumeration spanning multiple lines as an
example (in the latter example, the first parenthesis being outdent of
one character, the first item which comes next to it, has thus the
same indent as all the following items). But as this outdent is always
to be there, if the construct is not parenthesized, then the
indentation is not the expected one. So, when it happens a simple
expression has to stand to the next line, indented, just because it
would have otherwise go beyond the line width limit (which is 67
characters when I post example source on c.l.a), it turns to be a
simple expression with parenthesis.

That's true that a lot of parenthesized constructs of the Ada language
may seems ambiguous at first sight. But these potential ambiguities
are gracefully handled by some language rules intended for adjustment
and ambiguity removal, so that it is finally not ambiguous at all.