Re: On pragma Precondition etc.

[Georg Bauhaus <rm.dash-bauhaus@futureapps.de>]

Alex R. Mosteo schrieb:

> Piling more suggestions: I understand the point of using pragmas by Gnat at
> this stage, but if this is going to be standardized perhaps the syntax of a
> declaration could be changed. For example:
> 
> function Foo (X, Y : Something) return Whatever with
>   Precondition (Blah) and
>   Postcondition (Urgh);
> 
> So now it is all a indivisible whole and the OP concern is gone.
> 
> I haven't seen Eiffel code. How do they do this?

feature Foo(X, Y : SOMETHING): WHATEVER is
		-- short comment
	require
		Precondition_Name: Blah
	do
		...
	ensure
		Postcondition_Name: Urgh
	end

The Pre-/Postcondition_Name serves a purpose similar to the
message parameter of Ada's pragmas.)

Eiffel's "contract view" of the class then hides the statements
of Foo. So you get a "spec".

feature Foo(X, Y : SOMETHING): WHATEVER is
		-- short comment
	require
		Precondition_Name: Blah
	ensure
		Postcondition_Name: Urgh


In addition, Eiffel's Pre-/Postcondition have inheritance rules
for the conditions.  Another spot where Ada would to be more
general.


--
Georg Bauhaus
Y A Time Drain  http://www.9toX.de