Re: On pragma Precondition etc.

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On Tue, 29 Jul 2008 16:19:20 +0200, Georg Bauhaus wrote:

> On the other hand, overloading keywords is an invitation
> to make things more puzzling.

I don't see it that way, especially after Unicode identifiers were allowed.
You can legally use *any* reserved keyword as an identifier. Just replace
vowels 'o', 'a', 'i' with their Cyrillic equivalents and enjoy.

> One reason is that context
> is needed for disambiguation of the meaning of a keyword.

So what? 

> Another thing is, should conditions have parts? Possibly
> named parts?

No, I think that any condition shall be related to a [sub]type, which gives
the name of. I prefer named equivalence.

> The names do not add anything. However, when one of the pre-s
> is false, the name is seen in the trace, and in the corresponding
> EXCEPTION object.

Conditions shall not propagate exceptions in the same partition.
 
> For the Class Invariant, Randy Brukardt has mentioned
> 
>     for T'Constraint use Function_Name;

I don't want names of conditions contaminating the program name spaces,
like generics do. Semantically they belong to a different program. This
should be syntactically visible.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de