- From: Jacek Kopecky <jacek@systinet.com>
- Date: 23 Jun 2003 15:41:25 +0200
- To: Sanjiva Weerawarana <sanjiva@watson.ibm.com>
- Cc: David Booth <dbooth@w3.org>, WS-Description WG <www-ws-desc@w3.org>
- Message-Id: <1056375685.2275.38.camel@localhost>
Sanjiva, I do think that the figures are wrong because the statement "an interface has one or more URLs at which it is available (through bindings)" doesn't fit the statement "resource has one or more interfaces". The former statement takes into account just one interface whereas the latter talks about one resource and you shouldn't mix them. On the one hand, an interface has one or more bindings, a binding has one or more URLs, each URL has precisely one resource; in effect an interface has one or more resources, which is a statement of no practical value (AFAICS now). On the other hand, a resource has one or more URLs, a URL has precisely one binding, a binding has precisely one interface; in effect a resource has one or more interfaces (but URLs and bindings are between a resource and its interface). Surely, a resource has one or more interfaces to mess with it (hopefully my ascii-art gets through OK): +----------+ +-------------+ | resource | -----------+------------- | interface a | +----------+ | +-------------+ | +-------------+ +------------- | interface b | +-------------+ But a resource also has one or more URLs at which it is available. Each URL has an interface which it implements: +----------+ +-- URL1 -------- +-------------+ | resource | ---+----+-- URL2 -------- | interface a | +----------+ | +-- URL3 -------- +-------------+ | | +-------------+ +------- URL4 -------- | interface b | +-------------+ A URL has a binding which indicates the language in which it communicates the interface: +----------+ +-- URL1 ---- binding1 --- +-------------+ | resource | --+-- URL2 --+ | interface a | +----------+ +-- URL3 --+- binding2 --- +-------------+ This way, it's possible to convey in the image that multiple resources can share bindings and interfaces. Best regards, Jacek Kopecky Senior Architect Systinet Corporation http://www.systinet.com/ On Fri, 2003-06-20 at 13:31, Sanjiva Weerawarana wrote: > "David Booth" <dbooth@w3.org> writes: > > > > It occurs to me that figures 1-1 and 1-2 in our draft[1] that indicate > > "service" and "resource" are incorrect. They show the _interfaces_ > > pointing to the "resource", whereas the targetResource attribute allows > the > > WSDL _service_ to point to a resource. So the line should be drawn from > > the _service_ to the resource -- not from the _interface_ to the resource. > > First let's consider the picture without a circle saying "service." > Here the model we have: > > resource > has one or more interfaces to mess with it > interface > has one or more bindings to use it via > binding > has one or more URL at which its available > > What we call a <wsdl:service> now is effectively everything from > interface to the URLs. That is, the syntactic element really > captures everything from the URLs to the bindings to the interface. > The picture could be refined IMHO to have the circle be around > all of the things from resource up. Then the line connecting that > blob to resource is indeed what we capture with @targetResource > on <wsdl:service>. > > > Did I miss something? > > No but you caught something as usual :). > > Sanjiva. >
Received on Monday, 23 June 2003 09:41:46 UTC