W3C home > Mailing lists > Public > public-ws-policy@w3.org > May 2007

Re: Bug 4414 proposed - Relationship to Issues 4393 and 4292

From: Monica J. Martin <Monica.Martin@Sun.COM>
Date: Tue, 01 May 2007 16:36:57 -0700
To: David Orchard <dorchard@bea.com>
Cc: ws policy <public-ws-policy@w3.org>, Charlton Barreto <charlton_b@mac.com>, Charlton Barreto <cbarreto@adobe.com>
Message-id: <4637CF19.40709@sun.com>


>David Orchard wrote: Here's the revised text roughly agreed upon by myself, Maryann, Asir and
>reviewed but not agreed by Monica. 
>
mm1: To clarify, some questions were asked and, perhaps since there were 
unanswered, they can be so here. It appears you have moved in the 
direction of our questions (even if we didn't know about it).

>Note, it has been revised somewhat since all of them reviewed it so the fragile consensus may have fallen
>into a pile of sticks like so much dead wood ready to be kindling to
>warm the fires of our hearts.  I have not looked at the relationship
>between this text and the optional or ignorable sections.  I think
>perhaps this could be "swizzled" into those.
>  
>
mm1: This is at the heart of the questions we asked particularly given 
the discussion we had on Issue 4393 (and its predecessor 4292).

>If Company-X knows about the EndOfLife Policy assertion, it may or may
>not include that assertion in an alternative in the first version.
>
mm1: We had originally asked irrespective of the use of wsp:Ignorable 
marker and these modes, had you considered using wsp:Optional - one 
alternative with and the other without the assertion. It appears you 
have moved in that direction.

>If
>it does include the assertion, marks the assertion with an "ignorable"
>attribute, and provides no alternative without the
>assertion, then:
>- a client that does not know about the assertion and using lax
>intersection will produce an intersection.  
>- a client that does not know about the assertion and using strict
>intersection will not produce an intersection.  
>- a client that does know about the assertion and using strict or lax
>intersection
>will produce an intersection. 
>  
>
mm1: When two issues referenced were first discussed, we spoke about 
understandability characteristic associated with a policy assertion 
marked with the Ignorable attribute, we now reintroduce this concept. We 
approved this as the final text (given Glen Daniel's question):

    Regardless of the chosen intersection mode, ignorable assertions do
    not express any wire-level requirements on the behavior of consumers
    - in other words, a consumer could choose to ignore any such
    assertions that end up in the resulting policy after intersection,
    with no adverse effects on runtime interactions.

Dave, does this assumption still hold true with your premises herein or 
provide additional premises that should be documented?

For reference:
Issue 4393: http://www.w3.org/Bugs/Public/show_bug.cgi?id=4393
Issue 4292: http://www.w3.org/Bugs/Public/show_bug.cgi?id=4292
Section 3.4.1, Primer: 
http://dev.w3.org/cvsweb/~checkout~/2006/ws/policy/ws-policy-primer.html?content-type=text/html;%20charset=utf-8#strict-lax-policy-intersection

>If it does include the assertion, marks the assertion with an
>"ignorable" attribute, and provides an alternative without the
>assertion, then:
>
mm1: Maps to resolution that resulted in Section 2.8 in Primer where we 
talk about ignorable and optional.

>- a client that does or does not know about the assertion and using lax
>or strict
>intersection will produce an intersection.  
>
>If Company-X adds theEndOfLife policy assertion type to a subsequent
>Alternative and does not provide an alternative without the assertion then after the policy expression has been deployed/used the same
>algorithm holds true, notably that a client using
>strict mode that does not understand the assertion will not intersect
>with the alternative.   If CompanyX adds the EndOfLife policy assertion
>with an ignorable attribute to only one alternative, and leaves another
>alternative 
>without the assertion alltogether, then clients
>using strict mode who do not understand the EndOfLife assertion with the
>ignorable information will still be compatible with the alternative that
>does not contain the EndOfLife policy assertion as per the intersection
>rules and the server can return an error if the request is received
>after the expiry date.
>
mm1: Then, as asked before, are we then saying that the use of the 
Ignorable marker given understandability augments behavior (see above 
and be below)?

>If Company-X knows about the EndOfLife Policy assertion, it can
>guarantee that clients that know or don't know about the EndOfLife
>Policy Assertion can intersect under any mode by providing one
>alternative with the assertion, and another alternative without the
>assertion.   Clients that know about the EndOfLife Policy
>assertion and performing strict intersection can guarantee interaction
>with
>services that know or don't know about the EndOfLife Policy assertion by
>providing one alternative with the assertion and one alternative without
>
>the assertion (wsp:Optional="true").  Clients that know about the
>EndOfLife Policy
>assertion and performing lax intersection can guarantee interaction with
>services that know or don't know about the EndOfLife Policy assertion by
>providing one alternative with the assertion and marked with
>ignorable="true".
>
>Cheers,
>Dave
>
>  
>
Received on Tuesday, 1 May 2007 23:36:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:20:50 GMT