Re: procedural vs object oriented

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

On Thu, 27 Apr 2006 16:01:41 +0200, Jean-Pierre Rosen wrote:

> Dmitry A. Kazakov a �crit :

>> Well, no:
>> 
>> if Read (File) then
>>    Foo;
>> else
>>    Bar;
>> end if;
>> 
>> The uncertainty of a dispatching call is one of the context, exactly as in
>> the example above. Provided, that there is nothing uncertain in how
>> dispatching works or what potential targets do.
>> 
> Of course, you can assume that every dispatching call is equivalent to a 
> case statement over all possibly redefined primitives. That works well 
> for one level.
> 
> But if you consider that each called primitive may in turn redispatch 
> internally, you end up with a combinatorial explosion.

Ah, but redispatch is luckily not allowed in Ada. The programmer has to 
explicitly convert the type to the class. That should not slip through peer 
review.

[ I would even disallow it altogether. There is your trick for such 
things.]

> In theory, yes, the analysis can be performed. In practice, no.

Yes, but it is a sort of "postmortem" or C-ish approach. There is nothing 
fundamentally evil in dispatching if LSP problematic is supported. It is a 
question of the language and tools. Something like SPARK could change it. 

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