Re: "recursive" interface inheritance from Roberto Chinnici on 2003-05-01 (www-ws-desc@w3.org from May 2003)

From: Roberto Chinnici <Roberto.Chinnici@Sun.COM>
Date: Thu, 01 May 2003 16:22:31 -0700
To: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
Cc: www-ws-desc@w3.org
Message-id: <3EB1AC37.5080002@sun.com>

Savas Parastatidis wrote:
> Kevin,
> 
> 
>>The current editor draft of part 1 spec [1] reads like that extensions
> 
> can
> 
>>be layered, but it doesn't say anything about if an interface can
> 
> appear
> 
>>multiple times in the inheritance hierarchy, for example,  is it legal
> 
> for
> 
>>interface A to extend interface B and C, while B extends A,  C extends
> 
> B,
> 
>>and so on? For lack of good terms, I call it "recursive" inheritance.
>>
> 
> 
> I think you're referring to "cyclic" graphs in an inheritance tree.
> Actually, you don't need 3 interfaces to present the problem.
> 
> -> = extends
> 
> A->B->A (cyclic graph) not valid

Correct. In 2.4.1. we should say that the "extends" relationship between 
interfaces must be well-founded. Noncircularity follows.

Roberto

--
Roberto Chinnici
Web Technologies and Standards
Sun Microsystems, Inc.

Received on Thursday, 1 May 2003 19:23:42 UTC