- From: Doug Davis <dug@us.ibm.com>
- Date: Fri, 20 Feb 2009 21:01:00 -0500
- To: Geoff Bullen <Geoff.Bullen@microsoft.com>
- Cc: "public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>, "public-ws-resource-access-request@w3.org" <public-ws-resource-access-request@w3.org>
- Message-ID: <OF0E61F8F9.027B3D18-ON85257564.00060EFE-85257564.000B13FF@us.ibm.com>
There are two problems with this. If "no dialect" means "everything I can
see" we'd still need to define a dialect for this - not a big deal but,
like I said, "no dialect" should just be a short cut for something more
explicit/verbose.
But the real issue I have with this is that the definition of the MEX
dialect is a bit screwy. First it seems totally arbitrary. The dialect
URIs defined in MEX aren't really chosen for any particular reason other
than they're the list of dialects that the MEX authors just happen to
choose when they penned the spec. Not the most thoughtful approach. As
proof... what does the "policy" dialect mean? Policy by itself is
meaningless. It needs to be attached to something - like WSDL elements.
And what about the "policyAttachment" dialect? What metadata does that
return? So, out of the 5 dialects defined in MEX we'd only return 2 -
wsdl and xsd. But even then - what schema are we returning? The xsd of
the resource that would be returned by a Transfer Get()? The xsd of the
enum items? The xsd of the events if its an event source? All of these?
Some of them? This dialect feels a little like a better defined "whateva"
but still pretty useless since after removing all of the
meaningless/undefined URIs you're left with just WSDL anyway.
This however does raise another issue.... should people be forced to
define a dialect in order to get a bunch of metadata returned? For a
moment let's assume we defined MEX to mean "the dialects defined in the
MEX spec" - this one grouping is now well defined and 'special'. Sure
some other spec/profile could do the same thing but since we're talking
about a bootstrapping mechanism how do I know whether or not the other
side knows about this new special dialect URI? It seems we should allow
for people to define a grouping on the fly and be allowed to specify a
list of dialect URIs instead of just one. I think I've mentioned this
before but I think I'll finally get off my duff and actually open an issue
this time. :-) This would remove the need for a dialect that means "what
those crazy MEX author's thought was important even though more than half
of those dialects are meaningless" - which means we can go back to my
current proposal of having MEX/nodialect == "everything I'm allowed to
see".
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
Geoff Bullen <Geoff.Bullen@microsoft.com>
Sent by: public-ws-resource-access-request@w3.org
02/20/2009 07:23 PM
To
Doug Davis/Raleigh/IBM@IBMUS
cc
"public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>,
"public-ws-resource-access-request@w3.org"
<public-ws-resource-access-request@w3.org>
Subject
RE: Issue 6404 - proposal
Doug,
Trying to use your words to describe the two cases we propose, I get the
following:
1. ?no dialect? = "everything I'm allowed to see"
2. ?MEX dialect? = just the dialects defined in MEX
This way point 2 remains the same definition as it is now. It allows a
client to return only the MEX dialects if required (rather than all of
them), which can be a useful subset grouping. Point 1 allows the client
to return all dialects (including app specific ones). If there are no
application dialects then point 1 and point 2 will return the same thing.
Does that make sense?
--Geoff
From: Doug Davis [mailto:dug@us.ibm.com]
Sent: Monday, February 16, 2009 5:13 PM
To: Geoff Bullen
Cc: public-ws-resource-access@w3.org;
public-ws-resource-access-request@w3.org
Subject: RE: Issue 6404 - proposal
I think "no dialect" should be a shorthand for something and that
'something' needs to be some dialect. We can change it from the 'mex'
dialect to something else, but since we're talking about a bootstrapping
situation where we (as a client) are talking to a bit of an unknown entity
(which is why we're using mex in the first place), having the value of "no
dialect" be something random doesn't sound like an interoperable solution
- we'd be back to the 'whateva' case - which we determined is pretty
useless. I don't see any reason not to have "no dialect" == "the mex
dialect" and have the mex dialect mean "everything I'm allowed to see".
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
Geoff Bullen <Geoff.Bullen@microsoft.com>
Sent by: public-ws-resource-access-request@w3.org
02/16/2009 05:26 PM
To
Doug Davis/Raleigh/IBM@IBMUS
cc
"public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>,
"public-ws-resource-access-request@w3.org"
<public-ws-resource-access-request@w3.org>
Subject
RE: Issue 6404 - proposal
Hi Doug,
Our intent is slightly different here. We would prefer that returning
metadata associated with the dialect:
[Body]/mex:GetMetadata/mex:Dialect=http://www.w3.org/2009/02/ws-mex
remain consistent and ONLY ever return metadata associated with dialects
defined in the MEX specification.
The changes we suggest would only apply to the default case where no
dialect is specified.
In this case it would normally return the same as above, unless it has
been redefined by a profile to return something else, including Profile
specific metadata dialects.
Does that makes sense?
--Geoff
From: Doug Davis [mailto:dug@us.ibm.com]
Sent: Tuesday, February 10, 2009 11:48 AM
To: Geoff Bullen
Cc: public-ws-resource-access@w3.org;
public-ws-resource-access-request@w3.org
Subject: RE: Issue 6404 - proposal
Geoff,
Actually, the "default value" doesn't change - its the meaning of the MEX
dialect, no?
So, we really should be tweaking the other paragraph - the one starting
with "barring...". And doesn't that cover the possibility of someone else
further constraining it?
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
Geoff Bullen <Geoff.Bullen@microsoft.com>
Sent by: public-ws-resource-access-request@w3.org
02/10/2009 02:41 PM
To
Doug Davis/Raleigh/IBM@IBMUS, "public-ws-resource-access@w3.org"
<public-ws-resource-access@w3.org>
cc
Subject
RE: Issue 6404 - proposal
Doug,
It does not appear that the wording:
?When this element is not present, the implied value is the MEX dialect.?
correctly expresses the sentiment that we agreed too earlier. Can we
suggest using something more like:
?When this element is not present, the implied value is the MEX dialect.
However, the actual value may be defined by communities within the context
of particular application domains and could include application specific
metadata.?
--Geoff
From: public-ws-resource-access-request@w3.org
[mailto:public-ws-resource-access-request@w3.org] On Behalf Of Doug Davis
Sent: Sunday, February 08, 2009 5:49 PM
To: public-ws-resource-access@w3.org
Subject: Re: Issue 6404 - proposal
Resending since the html doesn't show up in the archives.
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
__________________
With no more chatter on this one... here's my proposal:
Define the absence of a Dialect to mean the MEX dialect - something like:
[Body]/mex:GetMetadata/mex:Dialect
When this element is present, the response MUST include only Metadata
Sections with the indicated dialect; if the receiver does not have any
Metadata Sections of the indicated dialect, the response MUST include zero
Metadata Sections. When this element is not present, the implied value is
the MEX dialect.
<delete> there is no implied value and so the response may include
Metadata Sections with any dialect. </delete>
And define the MEX dialect - add the following after the above text:
[Body]/mex:GetMetadata/mex:Dialect="http://www.w3.org/2009/02/ws-mex"
Barring some additional constraints, not defined by this
specification, specifying the MEX dialect in a GetMetadata request message
means that the service SHOULD return all available metadata formats that
this client is allowed to retrieve.
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
Doug Davis/Raleigh/IBM@IBMUS
Sent by: public-ws-resource-access-request@w3.org
01/29/2009 10:11 PM
To
Geoff Bullen <Geoff.Bullen@microsoft.com>
cc
"public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>,
public-ws-resource-access-request@w3.org
Subject
Re: Issue 6404 - use of "whatever"
Along those line, it would seem that saying something like "barring some
negotiation, the absence of a Dialect value is equivalent tousing the MEX
dialect". Gives the freedom for someone to profile it later - but
otherwise we make sure "null" is well defined.
thanks
-Doug
______________________________________________________
STSM | Standards Architect | IBM Software Group
(919) 254-6905 | IBM 444-6905 | dug@us.ibm.com
Geoff Bullen <Geoff.Bullen@microsoft.com>
Sent by: public-ws-resource-access-request@w3.org
01/29/2009 09:06 PM
To
"public-ws-resource-access@w3.org" <public-ws-resource-access@w3.org>
cc
Subject
Issue 6404 - use of "whatever"
This issue is about defining the MEX dialect and defining what gets
returned.
http://www.w3.org/Bugs/Public/show_bug.cgi?id=6404
In particular, I was asked to provide an example of why it might be
useful, in the case where no dialect is specified in the GetMetadata
request, for the service itself to be able to decide what it would return
(the so-called ?whatever? case). The other option would be for this case
to return all MEX sections.
The best example I can provide for the ?whatever? case is this:
If the MEX specification gets ?profiled? for a specific purpose, it would
be very useful to allow the profile to be able to specify what metadata is
to be returned in this default case (especially the non-MEX defined
metadata sections). If you do not do this then each profile would have to
define some separate dialect to mean ?give me all the metadata within my
profile?. Thus the default case gives you an over-loadable definition of
?all? or perhaps ?normal?, which can include non-MEX defined sections.
In a typical profiled case:
Nothing = ?return all metadata within my profile?
MEX = ?return all MEX dialects?
If it is not a profiled implementation, the spec could be recommend that
the implementation return:
Nothing = MEX = ?return all MEX dialects?
Received on Saturday, 21 February 2009 02:01:45 UTC