Re: Silly and stupid post‑condition or not ?

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

On 31.01.12 17:24, Dmitry A. Kazakov wrote:
> On Tue, 31 Jan 2012 16:34:23 +0100, Georg Bauhaus wrote:
>
>> The program below is also an example of Ada's POV.
>
> I fail to see your point.

The point is that from Ada's POV, there is a preference
for detecting many things early. But this preference does
not mean that all things would be detected by the compiler,
or should be detected by the compiler.
Ada provides for things checked at run-time, allowing
programs to be written that can only be written because
some amount of information need not be present at compile
time.

Ada does allow writing if statements without else part. Thus
suppose the argument is that a postcondition should either
state everything or, if this is not feasible, state nothing
at all (True). From Ada's POV, then,  nothing should ever
be implicit, or unspoken.  All if-statements should really
be case-statements without "when others". Though shalt not
write "if" without also stating what "else" should happen.

Not saying that deferring until run-time, or leaving out
the "else" part, is always a good idea, but given what
Ada is, the motto all-or-nothing is not Ada's POV, IMHO.
Hence, there is a mode of writing postcondition that lacks
rigor but still reduces the risk of errors, by reducing
their number, by informing the client programmer. Obviously,
there are uses of Ada where one wants
   Count (Number_of_Errors) in Boolean.
I cannot really think of a way the language can hope to do
more than encourage this relation.