Re: Ignorable assertions and interoperability

Expectations are different from the provider's perspective.
By marking the assertion as ignorable providers will see it being ignored by lax-mode consumers. This is what the providers expect. But then they might get some other consumers failing as they chose to ignore the provider's advice. 
Before (and now) normal assertions were/are not expected to be excluded  by consumers from the intersection...if consumers don't recognize them at the intersection time then they fail...

Cheers, Sergey


> Sergey, how is this ostensible interoperability problem any different 
> from the situation we had before the introduction of Ignorable, i.e. 
> where we had nothing but a strict intersection algorithm?
> 
> Fabian
> 
> 
> Christopher B Ferris wrote:
>>
>> I see now what you mean. However, IMO, this isn't an interoperability 
>> issue, as much as one of
>> understanding the reality that despite the fact that a policy provider 
>> might mark something as
>> ignorable, there will always be some policy consumers that will ignore 
>> that advice:-)
>>
>> I guess it is a matter of expectation setting.
>>
>> 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
>>
>> "Sergey Beryozkin" <sergey.beryozkin@iona.com> wrote on 02/21/2007 
>> 08:01:13 AM:
>>
>> > Hi Chris
>> >  
>> > "I guess I don't understand why strict mode presents interoperability
>> > challenges. We have both strict and lax mode intersection for a reason.
>> > Those policy consumers that don't want to ignore assertions that are
>> > marked as ignorable can use strict to achieve that objective. Those 
>> that
>> > are okay with ignoring what is marked ignorable can use lax mode. The
>> > policy consumer has the choice to do whatever they feel is right for
>> > their circumstances. "
>> >  
>> > I agree. From the consumer's perspective we have no issues at all.
>> > The things are slightly different from the provider's perspective 
>> though.
>> > Provider marks the assertion as wsp:ignorable=true so that it can be
>> > ignored for the intersection purposes. Otherwise why else would the
>> > provider do it ? If the provider wants the assertion be understood
>> > always then it would just expose that assertion as the normal
>> > required assertion.
>> > But the provider does not aware of what mode consumers will be
>> > using. By marking the assertion as ignorable the provider can get
>> > some consumers fail to consume the service if they work in the
>> > strict mode. Yes, that's what consumers chose to but I think it's
>> > not something a provider will really want... So it's a possible
>> > interop concern at the WS-Policy level
>> >  
>> > Do you see what I mean ? Would you agree ?
>> >  
>> > Cheers, Sergey
>> >  
>> >
>> > ----- Original Message -----
>> > From: Christopher B Ferris
>> > To: Sergey Beryozkin
>> > Cc: public-ws-policy@w3.org ; public-ws-policy-request@w3.org
>> > Sent: Wednesday, February 21, 2007 12:32 PM
>> > Subject: Re: Ignorable assertions and interoperability
>> >
>> >
>> > Sergey,
>> >
>> > Thanks for elaborating.
>> >
>> > Please see my inlined comments below.
>> >
>> > 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-policy-request@w3.org wrote on 02/21/2007 04:36:19 AM:
>> >
>> > > Hi Chris
>> > >  
>> > > Thanks for your comments.
>> > >  
>> > > I agree, I should've created a bug with a specific proposal rather
>> > > just suggesting (the editors :-)) to do some revisioning of the
>> > primer's text.
>> > >  
>> > > One goal of this email was to explain why I was concerned about the
>> > > interoperability statement during the concall as I promised at the
>> > > time to reply in email...
>> > > As far as wsp:ignorable and WS-Policy interoperability were
>> > > concerned, one possible take on it can be that using wsp:ignorable
>> > > might cause at the moment at least WS-Policy-level interoperability
>> > > problems due to third-party consumers using a strict mode. This is
>> > > one interop concern.
>> >
>> > I guess I don't understand why strict mode presents interoperability
>> > challenges. We have both strict and lax mode intersection for a reason.
>> > Those policy consumers that don't want to ignore assertions that are
>> > marked as ignorable can use strict to achieve that objective. Those 
>> that
>> > are okay with ignoring what is marked ignorable can use lax mode. The
>> > policy consumer has the choice to do whatever they feel is right for
>> > their circumstances.
>> >
>> > > As far as a provider is concerned, I believe a provider's motivation
>> > > to mark the assertion as wsp:ignorable is to try to reach with the
>> > > (assertion) message to as many requesters as possible and yet
>> > > continue to interoperate at the ws-policy level with ideally every
>> >
>> > Agreed.
>> >
>> > > requester out there. Thats's another possible view on what wsp:
>> > > ignorable means to the provider as far as a ws-level interop is 
>> concerned.
>> > > Then there's on the wire interoperability which is what was referred
>> > > to during the call.
>> >
>> > I still don't understand the interop concern.
>> >
>> > >  
>> > > Hopefully this explains the reason behind the message I've sent.
>> > >  
>> > > I've reviewed the primer and the guidelines yesterday and I've seen
>> > > just a few references to the interoperability term. As far as wsp:
>> > > ignorable and interop are concerned, section 2.7 adequately refers
>> > > to both on the wire interop and the ws-policy level interop
>> > > (implicitly by advising to be aware of the impact of this attribute
>> > > on the compatibility of policies).
>> > > I'll add a bug with a proposal to add a minor update to that section
>> > > (with respect to referring to interop). Specifically, I'll propose
>> > > to add a text sent by yourself earlier on the ignorability being at
>> > > the discretion of the requester.
>> > >  
>> > > Cheers, Sergey
>> > >  
>> > >  
>> > > ----- Original Message -----
>> > > From: Christopher B Ferris
>> > > To: Sergey Beryozkin
>> > > Cc: public-ws-policy@w3.org ; public-ws-policy-request@w3.org
>> > > Sent: Tuesday, February 20, 2007 12:52 PM
>> > > Subject: Re: Ignorable assertions and interoperability
>> > >
>> > >
>> > > Sergey,
>> > >
>> > > Would you please log this as a bug against the primer and guidelines
>> > > so that it can be tracked?
>> > >
>> > > Also, it would help to have specific areas of the primer and
>> > > guidelines that mention interoperability
>> > > so that we can focus on what exactly needs to be changed.
>> > >
>> > > Finally, if you could provide a proposal to address your concerns,
>> > > that would help greatly towards
>> > > closing the issue.
>> > >
>> > > 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-policy-request@w3.org wrote on 02/20/2007 05:52:22 AM:
>> > >
>> > > > Hi
>> > > >  
>> > > > During the latest concall it was recommended to advise not to use
>> > > > ignorable assertions if the interoperability would be affected...I
>> > > > thought it was a strong statement at a time.
>> > > > The reason for that was that I was assuming at a time a WS-Policy
>> > > > level interoperability was referred to.
>> > > > Most of the time it's obvious what interoperability the spec/primer
>> > > > texts refer to, but I feel it would be useful to revisit (in the
>> > > > primer and guidelines) all references to the 'interoperability'
>> > > > terms and qualify them as appropriate...
>> > > >  
>> > > > Cheers, Sergey Beryozkin 
>