- From: Doug Davis <dug@us.ibm.com>
- Date: Tue, 5 Jan 2010 13:03:22 -0500
- To: Gilbert Pilz <gilbert.pilz@oracle.com>
- Cc: ashok.malhotra@oracle.com, "public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>
- Message-ID: <OFB0343ED6.CFBCC270-ON852576A2.00631FA1-852576A2.0063344F@us.ibm.com>
Actually I think you can do it with schema. The definition of a
stand-alone mex:Location doesn't have to be the same xsd as the one under
a mex:MetadataSection.
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
The more I'm around some people, the more I like my dog.
Gilbert Pilz <gilbert.pilz@oracle.com>
01/05/2010 12:58 PM
To
Doug Davis/Raleigh/IBM@IBMUS
cc
ashok.malhotra@oracle.com, "public-ws-resource-access@w3.org"
<public-ws-resource-access@w3.org>
Subject
Re: Syntax proposal for Issue 7728
Fair enough, but these rules would not be enforceable via XML schema
validation (at least as far as my workman's understanding of XML schema
sees it) and would thus require programmatic steps to validate any
mex:Metadata document. So, in order to optimize the inclusion of metadata
references in EPRs to reduce the number of wrapping elements, I will now
be required to check that @Dialect, @Identifier, and @View don't appear as
attributes of either mex:MetadatReference or mex:Location when they appear
as children of mex:MetadataSection. Doesn't seem like a good trade-off to
me.
- gp
On 1/5/2010 9:30 AM, Doug Davis wrote:
We like rules ;-) The spec shouldn't allow for those attributes to
appear in both places and it depends on how its used. Within a
mex:metadatasection - no attributes on the location. Stand-alone, yes
mex:location has the attributes.
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
The more I'm around some people, the more I like my dog.
Gilbert Pilz <gilbert.pilz@oracle.com>
01/05/2010 12:11 PM
To
Doug Davis/Raleigh/IBM@IBMUS
cc
ashok.malhotra@oracle.com, "public-ws-resource-access@w3.org"
<public-ws-resource-access@w3.org>
Subject
Re: Syntax proposal for Issue 7728
I really don't like the idea of adding @Dialect, @Identifier, and @View
(people keep forgetting this, but it is important) to
mex:MetadataReference/mex:Location because, as I said before, you then
have to deal with (i.e. either specify the semantics for or rule out the
existence of) situations in which some or all of these attributes appear
at both the mex:MetadataSection "level" and the
mex:MetadatReference/mex:Location level.
For example, what does it mean if you have:
<mex:Metadata>
<mex:MetadataSection Dialect="wsdl">
<mex:Location Dialect="xsd">
or
<mex:Metadata>
<mex:MetadataSection Dialect="wsdl">
<mex:MetadataSection Identifier="someURI">
W/regards to "showing all 3", I think we need to be careful to distinguish
between what is allowed by WS-Addr extensibility and what we (WS-MEX)
actually define semantics for. (1) is allowed by WS-Addr, but I don't see
WS-MEX adding any value to this case.
- gp
On 12/29/2009 9:57 AM, Doug Davis wrote:
I tend to agree. Proposal #2 doesn't really add anything except a useless
wrapper. Proposal #1b might not work though since not all metadata under
the wsa:Metadata element could be "WSDL" (or whatever URI is in the
Dialect attribute). Proposal #1a, if adopted, should be expanded to
remove any ambiguity though. Since there are three ways that metadata can
be represented we should show all 3:
1: Inlined
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<wsdl:definitions ...> ... </wsdl:definitions>
</wsa:Metadata>
</wsa:ReplyTo>
2: URL Reference
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:Location Dialect="wsdl"> some-URI </mex:Location>
</wsa:Metadata>
</wsa:ReplyTo>
3: EPR Reference
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:MetadataReference Dialect="wsdl">
<wsa:Address> some-URI </wsa:Address>
</mex:MetadataReference>
</wsa:Metadata>
</wsa:ReplyTo>
However, we should also make it clear that there IS a case where the mex
wrappers might appear - its the case where the metadata itself is the MEX
document.
So, if doing a Transfer.Get() to some mex:MetadataReference returns:
<mex:Metadata>
<mex:MetadataSection Dialect="wsdl">
<wsdl:definitions> ... </wsdl:definitions>
</mex:MetadataSection>
</mex:Metadata>
Then when this metadata is inlined into an EPR it would appear like this:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:Metadata>
<mex:MetadataSection Dialect="wsdl">
<wsdl:definitions> ... </wsdl:definitions>
</mex:MetadataSection>
</mex:Metadata>
</wsa:Metadata>
</wsa:ReplyTo>
and the above EPR is semantically (from a metadata perspective) equivalent
to:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:MetadataReference Dialect="ws-mex">
<wsa:Address> some-URI </wsa:Address>
</mex:MetadataReference>
</wsa:Metadata>
</wsa:ReplyTo>
Notice the Dialect is "ws-mex", not "wsdl".
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
The more I'm around some people, the more I like my dog.
ashok malhotra <ashok.malhotra@oracle.com>
Sent by: public-ws-resource-access-request@w3.org
12/29/2009 11:50 AM
Please respond to
ashok.malhotra@oracle.com
To
"public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>
cc
Subject
Syntax proposal for Issue 7728
This proposal suggests a simplified syntax for attaching Metadata to MEX
GetMetadata. The proposal adds detail as a followup to the first point
in the original statement of this issue. The second point has been
addressed via other issues.
We discussed this on our last telcon. I'm following up with a specific
proposal.
The proposal for issue 6463 which we are close to accepting has three
levels of wrapping for the metadata, viz:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:Metadata>
<mex:MetadataSection Dialect="wsdl">
<mex:Location> http://ibm.com?wsdl </mex:Location>
</mex:MetadataSection>
</mex:Metadata>
</wsa:Metadata>
</wsa:ReplyTo>
We think that this is too many layers of wrapping and suggest that one
or two layers be removed.
Proposal 1
Remove both the <mex:Matadata> and the <nex:MetadataSection> wrappers.
The above example would then look like:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:Location Dialect="wsdl"> http://ibm.com?wsdl </mex:Location>
</wsa:Metadata>
</wsa:ReplyTo>
This syntax without any mex wrappers is already allowed by WS-Addressing.
Note that the Dialect attribute has been moved to the <mex:Location>
element. The Identifier attribute could be moved similarly.
Alternatively, extension facilities could be used to add these
attributes to the <was:Metadata> element. The above example would then
look like:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata Dialect="wsdl">
<mex:Location> http://ibm.com?wsdl </mex:Location>
</wsa:Metadata>
</wsa:ReplyTo>
Proposal 2
The above proposal does simplify the syntax but requires the Dialect and
Identifier attributes to be made available on other elements. If we
keep the <mex: MetadataSection> this is not necessary although the
proposal gets a bit more verbose:
<wsa:ReplyTo>
<wsa:Address> http://ibm.com </wsa:Address>
<wsa:Metadata>
<mex:MetadataSection Dialect="wsdl">
<mex:Location> http://ibm.com?wsdl </mex:Location>
</mex:MetadataSection
</wsa:Metadata>
</wsa:ReplyTo>
--
All the best, Ashok
Received on Tuesday, 5 January 2010 18:04:14 UTC