Re: Differentiating between Connectors and Objects

["Sameer Sahasrabuddhe" <sameerds gmail com>]

2008/7/11 Hans Breuer <hans breuer org>:

> I don't think this is the right thing to do for an expandable type system,
> i.e. I will not accept which moves knowledge about every object type name in
> existance into the core.

Which is why I thought that if it is really important, then it should
be in the vtable. Maybe not as "is_connector(obj)" but possibly as
'get_role(obj, "connector")' since that makes it extensible. But I am
just thinking aloud here ... I have not given any serious thought to
this.

> As tried to outline before not type is the key here but interface,
> respectively existance of properties.
>
> So for shared property definitions the proposed solution is fine, for object
> types it is not.

Ah, it seems I am out of my depth here. I am not really sure what
everyone is talking about, and I like it that way, unless it gets
close to some itch I have with Dia. Anyway, my post was triggered by
the ugliness in the pseudocode, with all those hard-coded string
comparisons. I just piped in to say that such ugliness should be
inside a nice function that can be updated whenever the implementation
details change. If the function also turns out to be useful for other
purposes, more power to it, then!

Sameer.
-- 
Research Scholar, Department of CSE, IIT Bombay
http://www.it.iitb.ac.in/~sameerds/