W3C home > Mailing lists > Public > www-ws-desc@w3.org > October 2002

Re: importing docs in the same namespace

From: Sanjiva Weerawarana <sanjiva@watson.ibm.com>
Date: Mon, 7 Oct 2002 22:24:42 +0600
Message-ID: <00c801c26e1e$0d318aa0$9a00a8c0@lankabook2>
To: "Amelia A Lewis" <alewis@tibco.com>
Cc: "WS-Desc WG \(Public\)" <www-ws-desc@w3.org>

"Amelia A Lewis" <alewis@tibco.com> writes:
> Umm, okay.  I had thought that the argument started from "it was
> possible in 1.1, so it should be possible in 1.2."  That is an entirely
> different animal than "it should be added to 1.2."
> My interpretation of the 1.1 specification was that it was not possible.

I disagree: It is absolutely possible with WSDL 1.1.

> Local discussion (web services geeks at TIBCO) tends toward the "sounds
> difficult to implement".  This is what got me interested: general
> consensus here was that it wasn't possible before, but the feature
> request was framed as "continue to support."

Why is it hard to implement? 

It *is* a continue to support a feature as well, but I was suggesting
that we consider it on its own merits. However, if it'll help folks
then its fine with me to consider it that way.

> I'm not much in favor of an ambiguous import mechanism.  

Me neither and I doubt anyone else is either.

> In fact, I
> would argue that the language of the spec should also forbid an import
> that has the same import/@namespace (=definitions/@targetNamespace) as
> the value of the containing definitions/@targetNamespace (that is, make
> the importing WSDL participate in the required uniqueness of
> namespaces).

I disagree.

Note that one CAN have multiple WSDL documents for the same
targetNamespace. That's a fundamental design feature of namespaces
that we aren't about to walk away from .. so I don't understand
the comment "required uniqueness of namespaces."

> I don't see any gain for permitting multiple instances of a key, when
> creating additional keys is easy.

I don't like to accept arguments based on something in the schema;
the schema should be written to support the language, not the
language to support the schema.

Do you agree with my scenario? Do you believe its ok to force
people to put Java classes into different Java packages as I 
described? The WSDL case is similar.

> I do think that the following ought to be completely clarified:
> can import/@namespace have a different value than the
> definitions/@targetNamespace of the imported WSDL?
> can any imported namespace match the importing WSDL's
> definitions/@targetNamespace?
> must import/@namespace be unique?
> My understanding of current status: no, no, yes.  After re-reading the
> 1.1 spec, I think that perhaps "unspecified, unspecified, yes" may
> reflect its state.

Sure, clarification is good, no matter which way decisions
are made.

Received on Monday, 7 October 2002 12:26:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:40 UTC