Class-Interface compatibility

[Frits van Bommel]

[fixed upside-down post]

Scott S. McCoy wrote:
> On Sun, 2008-03-30 at 23:25 +0200, Frank Benoit wrote:
>> COM is the only reason I am aware of, why interfaces are not implicit 
>> compatible to to object.Object. But COM is a very special case which 
>> probably can get a special solution?
>>
>> How about making /IUnknown/ a special interface without this 
>> compatibility and have all interfaces, not derived from IUnknown be 
>> compatible to Object?

> It is kind of curious why interface typed identifiers don't implicitly 
> cast to Object, and you must explicitly cast.
> 
> But regardless of the reason, adopting a convention (and a bad one at 
> that) such as prefixing interfaces with "I" probably shouldn't be done 
> at the language level.  And even more, D already has avoided creating 
> any kind of special types (you can throw Object, for instance, and it's 
> not a derivative of /Exception/) so far as I can tell...why start now?

He didn't propose handling interfaces starting with 'I' differently. He 
proposed handling interfaces derived from IUnknown differently. They're 
already treated differently[1], but he wanted to change the behavior of 
other interfaces without adding the same change to IUnknown-derived 
interfaces (since it wouldn't work there[2]).


[1]: See the last part of <http://www.digitalmars.com/d/1.0/interface.html>.

[2]: Objects referred to by COM interface references (see [1]) may be 
implemented in a different language than D, and so may not be compatible 
with the D Object class at all.