RE: First cut policy write up

I stand humbly corrected . . .
 
- gp


  _____  

From: Christopher B Ferris [mailto:chrisfer@us.ibm.com] 
Sent: Monday, December 04, 2006 12:25 PM
To: Gilbert Pilz
Cc: David Illsley; public-ws-addressing@w3.org;
public-ws-addressing-request@w3.org; Yalcinalp, Umit
Subject: RE: First cut policy write up



I continue to be fascinated with this thread. 

> As I recall, we decided that "UsingAddressing" always has one meaning
which
> is that WS-Addr is "supported". 

If this is the case, and that when in conjunction with wsdl:required='true'
that it effectively changes the 
semantic to that of the 11th commandment: "Thou shalt use WS-Addressing",
then I guess I am confused, 
because wsa:UsingAddressing no longer has the semantic of "WS-Addressing is
supported" except implicitly. 
(If the endpoint is admonishing its prospective consumers that they MUST use
WS-Addressing, lest they 
suffer the Wrath of Khan, or the WS-Overlords, or whomever dishes out wrath
in the context of WS-*dom, 
then that admonishment obviously implies that WS-Addressing is supported at
that endpoint, or else they would not 
have written that commandment into their WSDL tablet in the first place.) 
. 
The semantic of wsdl:required='true' is that it means that the consumer of
the WSDL MUST 
understand that EXTENSION in order that they be able to effectively use the
WSDL description [1]. WSDL2.0 
basically has the same semanic. wsdl20:required='true' effectively means
that the extension may change the 
semantic of the parent WSDL element in a manner in which if not understood,
would prevent the client 
from correctly interacting with the endpoint described. 

So, in both cases, what the semantic of wsdl:required='true' adds to a given
extension is that you MUST 
understand that extension's semantic. If wsa:UsingAddressing means
"WS-Addressing is supported", then 
the mere presence of wsdl:required='true' cannot change that semantic into
"Thou shalt use WS-Addressing". 
wsa:UsingAddressing still means "WS-Addressing is supported". 

If you want to convey a requirement that WS-Addressing MUST be used, then
you need an extension that 
says that,unambiguously in its semantic, absent the presenceor absenceof
"wsdl:required='true'. 

If you want the semantic that WS-Addressing is supported, but not required
(e.g. optional) then use wsdl:required='false' 
(the default in both cases IIRC) as a means of providing the consumer of the
WSDL to choose whether they 
want to "understand" the semantic of the wsa:UsingAddressing WSDL extension
(of course, I would strongly 
suggest renaming the extension to something a little more descriptive of the
semantic carried: e.g. wsa:UseAddressing). 

If the semantic of the extension means: "Thou shalt use WS-Addressing", then
clearly, it can be used in WS-Policy 
with the wsp:Optional='true' to indicate that there are two policy
alternatives available to the consumer: one with, 
and one without the admonition that "Thou shalt use WS-Addressing". 

IMO, wsa:UsingAddressing is broken if its current semantic is "WS-Addressing
supported" because you 
can't change its semantic with wsdl:required='true'since
wsdl:required='true' isn't intended to CHANGE the semantic 
of the extension, but merely to signal to the WSDL processor that it MUST
understand the extension's semantic. 

[1] http://www.w3.org/TR/wsdl#_language 
[2] http://www.w3.org/TR/wsdl20/#language-extensibility 

Cheers, 

Christopher Ferris
STSM, Software Group Standards Strategy
email: chrisfer@us.ibm.com
blog: http://www.ibm.com/developerworks/blogs/page/chrisferris
phone: +1 508 377 9295 

public-ws-addressing-request@w3.org wrote on 12/04/2006 02:15:07 PM:

> David's understanding of the semantics of the wsaw:UsingAddressing policy
> assertion is in line with my intentions in describing it and (I hope) in
> line with what the group decided on the 11/27 concall.
> 
> As I recall, we decided that "UsingAddressing" always has one meaning
which
> is that WS-Addr is "supported". When used as a WSDL extension in
conjunction
> with 'wsdl:required="true"' it means that WS-Addr is required.
Unfortunately
> there doesn't seem to be a way of both keeping the semantics of
> "UsingAddressing" constant across its use as a WSDL extension and a policy
> assertion *and* allowing the WS-Policy variation to express that the use
of
> WS-Addr is required. I thought we had discussed this and agreed that we
> would have to live with it. If you really need to express the fact that
> WS-Addr is required for an endpoint you will have to use
> wsaw:UsingAddressing as a WSDL extension and mark it with
> 'wsdl:required="true"'.
> 
> As for nested policy assertions, again, I thought we had discussed this
and
> decided that they were too complicated. I now think we were a bit
premature
> in that decision and I agree that nested policy assertions work best for
> expressing the relationship between the broad "UsingAddressing" assertion
> and the narrower "AnonymousReponses" and "NonAnonymousResponse".
> 
> - gp
> 
> "too forward and one back
>  blind fingers groping for the right track"
>  
> 
> > -----Original Message-----
> > From: David Illsley [mailto:david.illsley@uk.ibm.com] 
> > Sent: Monday, December 04, 2006 4:40 AM
> > To: Yalcinalp, Umit
> > Cc: Gilbert Pilz; public-ws-addressing@w3.org
> > Subject: RE: First cut policy write up
> > 
> > Comments below.
> > 
> > public-ws-addressing-request@w3.org wrote on 12/01/2006 10:34:05 PM:
> > 
> > > > I spent a while yesterday going over this proposal with 
> > Katy, Paco, 
> > > > and our WS-Policy development team and we have a couple 
> > of concerns.
> > > > 
> > > > 1. There is no way to mandate addressing in this proposal i.e. In 
> > > > normal form (once the wsp:Optionals etc have been expanded) the 
> > > > presence of wsaw:UsingAddressing only indicates addressing is 
> > > > supported.
> > > > We need a way
> > > > to say addressing is required. I don't have a proposal 
> > yet to deal 
> > > > with this.
> > > > 
> > > 
> > > I am really not following this point. Could you clarify? 
> > > 
> > > If you do not use wsp:optional and use the standard attachment 
> > > mechanisms, why wouldn't WS-Addressing be NOT required.
> > > 
> > > IF there is no alternative in the policy, the intersection 
> > algorithm 
> > > and thus the client will treat WS-Addressing assertion as 
> > an addition 
> > > that it needs to understood and thus make behavior required.
> > > 
> > 
> > I agree that in those circumstances, the UsingAddressing 
> > assertion would be required for the client.
> > However, the example states:
> > > > <wsp:Policy>
> > > >   <wsaw:UsingAddressing>
> > > > </wsp:Policy>
> > > > This policy indicates that the subject supports the use of
> > WS-Addressing. 
> > 
> > It explicitly does not say that inclusion of UsingAddressing in an 
> > alternative mandates the use of WS-Addressing, merely that it is 
> > supported, hence the concern.
> > 
> > David
> >