Re: Dummy

[kilgallen@eisner.decus.org (Larry Kilgallen)]

In article <t7zoti1cd7.fsf@calumny.jyacc.com>, Hyman Rosen <hymie@prolifics.com> writes:
> kilgallen@eisner.decus.org (Larry Kilgallen) writes:
>> To me "perfectly safe" does not allow "fail at runtime".
> 
> Nonsense. Ada has very similar situations. For example, suppose you
> have a dispatching function with more than one parameter of the
> controlling type. Ada will conduct a runtime check to make sure that
> all the controlling parameters have the same derived type when the
> call is made.

My comment was about the term "perfectly safe".
Your response was about the capabilities of Ada.
Am I missing the post of a third person who characterized Ada
as "perfectly safe" ?

Certainly Ada is not "perfectly safe", or Spark would not exist.
(I am not claiming that Spark is "perfectly safe".)

> "Perfectly safe" means that either the operation succeeds, or you
> are informed that the operation failed.

I have not read that as a formal definition in Ada books,
nor read it in this newsgroup.  I took it as ordinary use
of English.

Larry Kilgallen