Some comments (was Back to the issue... (was RE: Squeak codin
g style...))
Oca Emilio
eoca at afip.gov.ar
Wed Mar 3 16:32:07 UTC 2004
Wolfgang, all,
I think that Bobby Wolff first presented in a paper or
article this idea of two kinds of comments in every
method. One to explain method semantic, the other to
explain method code.
The first one was a must and the second was not
desired, but may be inevitable.
He paid especial attention on building polymorphic
hierarchies by this technique.
> Alejandro,
> I think I got your point about messages vs. methods.
>
> But I believe that assuming that two methods with the same
> selector implement the same message (semantics) does not hold,
> and is hard to enforce.
Yes, you cannot enforce that everywhere. But, within
a polymorphic hierarchy, I think is a must. If it cannot
be put the same semantic comment to an inherited method;
then that method is not compliant with its super and can
not be used within that hierarchy in a polimorphic way.
Wolff goes further and suggests that the comment should
be "See super implementor." based on a series of facts
that enumerates on his paper.
Another 2c.
Regards.
Emilio
>
> Furthermore, a method comment could not only contain
> the message comment, but also information about the
> particular implementation of that message.
>
> Just my 2c...
> Wolfgang
>
>
>
>
>
>
More information about the Squeak-dev
mailing list
|