Re: Discussion on issue 3599

Hi Ashok, Paul,

I have made the "reply" list smaller (since everybody in this discussion
is on the public list) and will reply to your summary and Paul's
questions in this mail.

Ashok Malhotra wrote:
> Paul:
> Let me summarize my current thinking.  Felix and Phillipe will tell us whether they agree.
> 
> 1. WSDL 1.1 documents are in XML format and have a URI.
correct.
> 2. Various aspects of the WSDL definition set contained in the document 
> are represented as XML elements.
correct. IMO the question is whether you think that these aspects are
enough for your purposes (see below).
> 3. XPath can be used to identify individual elements.  These elements can then
> be used as Policy Subjects using a Domain Expression in the AppliesTo in PolicyAttachment.
> 
> This does not address the import/export question.  
> Nor does it address the exact XPath syntax to be used.
> 
> My observation wrt XPointer is that the syntax is very similar to XPath
correct. I mentioned XPath to make clear that IMO we only should
identify individual elements. Of course you can do that with XPointer as
well, but you could do more (see below).
> except for
> the shorthand that can be used with properties defined as IDs (barenames).
as long as we only operate on elements and not on (yet to be defined)
components, that would be fine with me.
Some replies to Pauls mail below.

> 
> All the best, Ashok
>  
> 
>> -----Original Message-----
>> From: public-ws-policy-request@w3.org 
>> [mailto:public-ws-policy-request@w3.org] On Behalf Of Paul Cotton
>> Sent: Thursday, October 05, 2006 9:16 AM
>> To: Felix Sasaki; Philippe Le Hegaret; Ashok Malhotra
>> Cc: public-ws-policy@w3.org
>> Subject: RE: Discussion on issue 3599
>>
>>
>> Thank you for starting this thread.  Unfortunately I have 
>> several questions about your synopsis of your offline 
>> dialogue with Ashok and Philippe.
>>
>>> Philippe and me raised a concern about the XPointer 
>> mechanism: it points to WSDL 2.0 components, and not the XML 
>> representation of the WSDL document.
>>
>> I am confused by this statement.  Wouldn't a use of XPointer 
>> have to point to the XML representation since it has no 
>> knowledge of anything else?  Can you explain exactly what you 
>> mean by the "XPointer mechanism"?

I meant "an XPointer mechanism used to point to components", *if* that
is what Ashok has in mind. This is what I was (am) not sure about.

The XPointer framework allows using shorthand pointers to point to one
element in the information set, or scheme based pointers. The XPointer
registry at http://www.w3.org/2005/04/xpointer-schemes/ describes
registered schemes. There are e.g. also registered schemes to point to
WSDL 2.0 components, but of course nothing specific for WSDL 1.1.

So I would ask Ashok: do you just want to use the short hand pointers,
or an XPointer scheme? If you want to use an XPointer scheme, should it
be an existing, registered scheme (which one?) or a new one?

>>
>>> For WSDL 1.1., there is no clear definition of a component model.
>> Agreed.
>>
>>> We argued that using XPath instead of XPointer for the WSDL 1.1.
>>> external attachment would make it clear that we refer to the XML 
>>> representation, and there would be no need to define a 
>> component model 
>>> for WSDL 1.1.
>> See my above questions.

See my answer above. I wanted to get clearness from Ashok what he wants
to use XPointer for. The resolution at
http://www.w3.org/2006/09/13-ws-policy-minutes.html says only "Ashok
agreed to consider using the XPointer framework instead of trying to
define a fragment identifier for WSDL 1.1 resources.", but it is not
clear about if an XPointer scheme should be used, and which scheme to use.

>>
>>> One example difference between the component model in WSDL 
>> 2.0 and the 
>>> XML representation is that the former takes import/include into 
>>> account, while the latter doesn't.
>> This is an extremely important point that has not been made 
>> before in the WG discussion.

Correct. IMO, if the idea is to use XPointer shorthand pointers or one
of the existing registered schemes, we will be stuck with that problem.
But maybe that is o.k. to resolve the issue 3599?

>>
>>> Ashok had pointed out that there is not much difference between the 
>>> XPath and XPointer syntax and proposed to use the syntax he had 
>>> proposed for referring to WSDL 1.1 components [5] (search for "For 
>>> WSDL 1.1. we suggest").
>>> [5] 
>>>
>> http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0019.html
>>
>> The F2F meeting pointed out that using this syntax in [5] 
>> combined with the URI of the WSDL file and separated by a 
>> hash would look like a fragment identifier which is NOT 
>> feasible since there is no MIME type that defines this for 
>> WSDL 1.1.  Is Ashok attempting to revert back to that proposal?

I don't think so, I was only unclear about the usage of XPointer (see
above).

Felix

>>
>>> Using this syntax would IMO look like operating on the (yet to be 
>>> defined) WSDL 1.1 component model and not on the XML representation.
>> I agree that without a WSDL 1.1 component model then any 
>> approach is problematic unless there is significant work in 
>> the definition of the selected pointer mechanism to specify 
>> exactly what part of the WSDL file was being selected.  This 
>> sounds like doing the work to define a component model for 
>> WSDL 1.1 which is clearly out of scope of our work.
>>
>> /paulc
>>
>>
>> Paul Cotton, Microsoft Canada
>> 17 Eleanor Drive, Ottawa, Ontario K2E 6A3
>> Tel: (613) 225-5445 Fax: (425) 936-7329
>> mailto:Paul.Cotton@microsoft.com
>>
>>
>>
>>
>>
>>> -----Original Message-----
>>> From: public-ws-policy-request@w3.org [mailto:public-ws-policy- 
>>> request@w3.org] On Behalf Of Felix Sasaki
>>> Sent: October 5, 2006 1:44 AM
>>> To: public-ws-policy@w3.org
>>> Subject: Discussion on issue 3599
>>>
>>>
>>> This is my action item 121 "Felix to Start a mail thread on the 
>>> discussion between Ashok, PLH and Felix on issue 3599"
>>> http://www.w3.org/2006/10/04-ws-policy-minutes.html#action01 .
>>>
>>> Background: At the September f2f I described several issues 
>> as "out of 
>>> scope", see [1]. Issue 3599 [2] was one of them.
>>> After I had left the f2f early, this issue was discussed [3]. The 
>>> proposal from Ashok, to use fragment IDs for the external 
>> attachment, 
>>> brought up the problem that there is no mime type for WSDL 1.1. 
>>> defined, and it is unclear who would define it or what to 
>> do with the 
>>> existing WSDL 1.1. documents that don't have a mime type.
>>> Using XPointer to refer to WSDL definitions would solve the 
>> mime type 
>>> problem. Ashok had the AI to consider this proposal [4].
>>>
>>> Discussion between Ashok, PLH and Felix: Philippe and me raised a 
>>> concern about the XPointer mechanism: it points to WSDL 2.0 
>>> components, and not the XML representation of the WSDL 
>> document. For 
>>> WSDL 1.1., there is no clear definition of a component model. We 
>>> argued that using XPath instead of XPointer for the WSDL 
>> 1.1. external 
>>> attachment would make it clear that we refer to the XML 
>>> representation, and there would be no need to define a 
>> component model for WSDL 1.1.
>>> One example difference between the component model in WSDL 
>> 2.0 and the 
>>> XML representation is that the former takes import/include into 
>>> account, while the latter doesn't.
>>> Ashok had pointed out that there is not much difference between the 
>>> XPath and XPointer syntax and proposed to use the syntax he had 
>>> proposed for referring to WSDL 1.1 components [5] (search for "For 
>>> WSDL 1.1. we suggest"). Using this syntax would IMO look like 
>>> operating on the (yet to be defined) WSDL 1.1 component 
>> model and not on the XML representation.
>>> Felix
>>>
>>>
>>> [1] 
>>>
>> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0064.html
>>> [2] http://www.w3.org/Bugs/Public/show_bug.cgi?id=3599
>>> [3] http://www.w3.org/2006/09/13-ws-policy-minutes.html#item13
>>> [4] http://www.w3.org/2006/09/13-ws-policy-minutes.html#action05
>>> [5] 
>>>
>> http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0019.html
>>
>>
>>
>>
> 
>

Received on Friday, 6 October 2006 06:48:59 UTC