- From: Jacek Kopecky <jacek@systinet.com>
- Date: 17 Oct 2002 00:07:00 +0200
- To: Henrik Frystyk Nielsen <henrikn@microsoft.com>
- Cc: Martin Gudgin <mgudgin@microsoft.com>, Noah Mendelsohn <noah_mendelsohn@us.ibm.com>, XMLP Dist App <xml-dist-app@w3.org>
Henrik, please see inline. Jacek Kopecky Senior Architect, Systinet Corporation http://www.systinet.com/ On Wed, 2002-10-16 at 23:44, Henrik Frystyk Nielsen wrote: > > Jacek, > > I am not aware of us ever having said anything about whether there is > one or many nodes that may act in a single role. Unless we are > introducing such semantics for roles in general (which I would not be > happy about) then I don't understand why you separate the scenarios of > the node acting in a role being special from the ones where any > subsequent nodes acting in that role. I'm unsure about whether you understood me as saying there is always only one node playing a given custom role (which I haven't even thought) or as saying there is always many nodes playing a given custom role (which is impossible on short paths). I know that with the exception of SOAP-defined role names, any node can play any role. > Settling on a default for what to do will make the other case harder. > Leaving out the possibility of adding an attribute, the point of my > scenario was that the relay role AND the current default would allow > both cases while maintaining the optionality. However, I can't see this > being the case if the default was turned the other way. My table shows that flipped default handles more of the 6 targeting scenarios - basically five well and one partially (number 1) - whereas the special role only handles one scenario well (number 1) and two partially (numbers 3 and 6), leaving three scenarios impossible. My text later after the table shows why I give more weight to scenario 2 than to 1 - because of my perceived semantics of role names and their role in a system design. Btw, why are you leaving out the possibility of adding an attribute? I haven't noticed a message where you'd say you can't live with this solution (but then I haven't been following the thread in its completeness). If anything in my message is wrong (missing scenarios, wrong values in the table or the intended semantics of a role name), please say so explicitly because otherwise I might not catch it. Jacek
Received on Wednesday, 16 October 2002 18:07:31 UTC