- From: Jonathan Marsh <jmarsh@microsoft.com>
- Date: Wed, 26 Jun 2002 12:20:49 -0700
- To: "Martin Gudgin" <mgudgin@microsoft.com>, <www-ws-desc@w3.org>
Given that relative URIs are undefined for use as XML namespace URIs, should we mandate that this URI be absolute instead of merely non-zero length? > -----Original Message----- > From: Martin Gudgin [mailto:mgudgin@microsoft.com] > Sent: Tuesday, June 25, 2002 3:08 AM > To: www-ws-desc@w3.org > Subject: Target namespace in WSDL > > > Both WSDL and XML Schema have the notion of a target namespace. In XML > Schema this property can be absent, denoting constructs that are not > affiliated with a particular namespace. This was necessary because XML > Schema describes XML instances, and elements in an XML instance may be > unqualified ( not affiliated with a namespace ). > > The question I would like to pose is 'Does it makes sense to allow the > target namespace property of WSDL components to be absent?'. And I will > argue that it does not. WSDL does not describe XML instances, it > describes messages, portTypes, bindings and services. I think it makes > sense to mandate that these contructs always be affiliated with a > namespace. > > To this end, I propose that we mandate the 'targetNamespace' AII on the > definitions EII. And that we modify the spec to say that the value of > the AII must be a non-zero length URI. > > Changes to spec > > Section 3.1 > > Add a bullet between current bullets 2 and 3; > > * A target namespace attribute information item amongst its [attributes] > as described below > > > Section 3.1.1 > > Add a third bullet > > * A type of anyURI in the http://www.w3.org/2001/XMLSchema namespace > > Add prose > > The value of the targetNamespace attribute information item MUST NOT be > the empty string. > > > > Gudge
Received on Wednesday, 26 June 2002 15:21:21 UTC