Web Services Resource Access Working Group Teleconference

28 Jan 2010


See also: IRC log


Alessio Soldano, Red Hat
Ashok Malhotra, Oracle Corp.
Asir Vedamuthu, Microsoft Corp.
Bob Freund, Hitachi, Ltd.
David Snelling, Fujitsu, Ltd.
Doug Davis, IBM
Fred Maciel, Hitachi, Ltd.
Gilbert Pilz, Oracle Corp.
Jeff Mischkinsky, Oracle Corp.
Li Li, Avaya Communications
Martin Chapman, Oracle Corp.
Ram Jeyaraman, Microsoft Corp.
Sreedhara Narayanaswamy, CA
Tom Rutt, Fujitsu, Ltd.
Vikas Varma, Software AG
Wu Chou, Avaya Communications
Yves Lafon, W3C/ERCIM
Bob Natale, MITRE Corp.
Katy Warr, IBM
Mark Little, Red Hat
Paul Fremantle, WSO2
Paul Nolan, IBM
Prasad Yendluri, Software AG
Bob Freund, Hitachi, Ltd.
Sreedhara Narayanaswamy


<trackbot> Date: 28 January 2010

<Bob> scribenick: Sreed


Bob: I haven't seen any new progress on this

Ram: Eventing - event sync that there will no notifications, model eventing it is push - in case of enumeration consumer pulls

dug: Enumering over ever going queue - waiting for this to pop up

DaveS: first enumeration need to be generated - waiting
... submitting jobs (current list of jobs) that there is no jobs in the queue

dug: not necessary empty

Ram: can't process fault enumeration something got wrong with the fliter

gpilz: there is difference b/w badly constructed filter and approriate filter

Asir: will this condition ever occur?

RESOLUTION: 8157 is resolved as proposed - fault defintion will be modified approriately for the enumeration



<dug> In the case of a Put operation if the XPath expression selects more than one node, the implementation MUST return only the first selected node

<dug> In the case of a Put operation if the XPath expression selects more than one node, the implementation MUST process only the first selected node

Ram: previously XPath 1.0 can evaluate multiple paths dont use Put

RESOLUTION: 8191 resolved with comment nr. 5 in the issue description

8180, 8299, 8302

<gpilz> http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0190.html

gpilz: Reviewed the original proposal - questions how extended WS standardad, WST representation element looked into the optionality. First change WST get response

dug: representation element becomes the complete representation - lloking at the element get response or it is different it is wst representation

asir: I didnt see the conversation having benefits two different wrappers

gpilz: wst representation extend authors they need can use it if not required

Ram: I want to establish the wrapper should be consistent it should serve the purpose for all the operations

gpilz: it is the representaiton of the resource

Ram: can it is be a partial representation
... people would start inventing their own wrappers

gpilz: most reuse the wst specific to wrappers

gpiliz: change is in wst:Put a representation allowing for an idea dialects - element must be present - base behaviour no dialects - extension authors to resue it or not

Ram: when we do a wst:put full reprsentation now if choose fragment approach in that case I might not be using this representation it means to me representation it is used in full representation

gpilz: wst:PutResponse can get representation element back - it is optional
... samething for create & creatersponse didnt change
... in case if there is no dialect - extension of base dialect & there is dialect it shouldnt be

Ram: wst:put second sentense it is fine - assume there are no dialiects working on extensions - just a clarification

gpilz: yes

Ram: If i dont use dialects/extensions can I pass instruction in this representation element

DaveS: put & create are different in this context

Ram: my observation the represntation wrapper should contain the full representation

dug: we need to have the word "full"

Ram: use a dialect then will not have an element at all wrapper
... same applies to extension
... option send or not send back - question how as I client two possiblities get back the representation if I dont get back the repsentation about the update or create operation

gpilz: client doesnt draw the semantic conclusion

Ram: I send representation A & put response A the case is when you dont send me back B what do I conslude as what happened

gpilz: two case, I do a put & get back the representation can I conclude is there a semantic different be/w two - I say no, it might be service can use the response can do any kind of conslusion you can do implict get

Ram: Clinet <-> Service, possiblities get back A, get back B (modified) or nothing the modification proposed case-A get back A will know the representation

Bob: might not the representation in this case

DaveS: B is the same case

dug: easier to jump to the end what sentense which requires
... everybody understand this
... what sentense is required

Ram: In the case of nothing what I prefere the client to do in this case what ever is suppplied the service didnt make any modidications in the caseB I am not worried

Bob: all the 3 response will have same information
... current spec - recieveinng A, recievingB & nothing returns the same

Ram: when there is noting client side the service didnt make any change

Bob: Some thing change the request or something different then asked

Ram: Current model send an a, get back b what specs say send an a service knows as a - dont need to repeat -- client know about this

Bob: what is means it sends back b
... is b really what it is

<dug> Absence of this element can be interpreted as the update request was successfully processed in its entirety (assuming no intervening mutation operations are performed).

Ram: Client needs a snapshot to work with

dug: I asked for something & posted

dug; is this what should be in spec

<dug> Absence of this element implies that the updated representation does not differ from the supplied representation (assuming no intervening mutation operations are performed).

Bob: question - about this protocol I presuming the changes might occur processing the state some what done by the application, we have transfer proptocol using this protocol in some application - possible change in the applciation it is transfer itself? what is causing the resoucre representioon different
... concurrent changes at any time - false sense of security giving to client - what it did work - service filling buffer by the time client gets it message (nano seconds) - instance that client recieves the date it could be totally unreliable
... suggesting if you take the model response gets back extacly the way it is or diffrerent particular issue mention about the concurrency - put response in this situation could be harmful

Ram: Using the concurrency argument - I need to know what is do scope at the particiular request

Bob: Processing for all the requests

dug: there is a disconnect - what you get an put response & I think ram is saying on the data is represnted is accurate

Bob: I am sggesting far better than doing that

gpilz: know about the resource as what should do is get

DaveS: data get back will not tell anything as a service I have infact there is no way send you back data where as completly open 7GB to send back there are so many cases cant be sure that empty representation sending hugh amount of data.

Ram: Clarification send something than change the time stamp

DaveS: anything changes

gpilz: represenattion is a full representation

Ram: whether the services are obligated to ship back how do as I client what expectations not seeing any response back at the response

DaveS: many of the errors might havent happened

gpilz: specs says something about this

<MartinC> unless its transactional;)

<MartinC> ws-tx?

<MartinC> +1 to jeff - distributed systems 101

Ram: I am fine with gpilz proposal
... are there any schema changes

dug; I will take care of it

RESOLUTION: resolved 8180, 8299, 8302 as proposed in http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0190.html


RESOLUTION: 8273 is deferred until after last call


gpilz: I dont have a proposal for this

Asir: is this related to 8196

Bob: combine 8196 & 8229

gpilz: they might be different
... to use qualified names - request when evaluted resource would produce different results

Ashok: what is the problem

DaveS: mandate what we have full namespace - full qualified

gpilz: what is the first sentense means

<asir> I like Yves' suggestion!

Tom_Rutt: *: means full namespace so this would be approriate

gpilz: as clinet I need to know that before consutructing
... what namespace bindings are significant means even I know the schema should have elements only having determination to do get different namespace I cant really evaluate need to have complete snapshot

dug: last sentense we get what is required if that is ok then go for it

Bob: how do we fix the second sentense

gpilz: namespace bindling are significant for all the elements

<Tom_Rutt> If I do an xpath evaluation in an xml tool, if the expressions uses element names which are not namespace qualified, they will only work agains an input document which has that namespace as the default namespace

Bob: the namespace is document in XPath - suggestion

Tom_Rut: XPath has qualified when I am writing namespace prefix

DaveS: Putting some not qualified Tom said it is going to match default namespace I am happy with it

RESOLUTION: 8229 resolved by removing the text in question


Bob: there is no proposal for this

gpilz: I don;t understand it

Bob: any proposal to clarify

RESOLUTION: 8306 is deferred until after last call


gpilz: have we changes the modes siginficantly

dug: we havent decided yet
... I need to work with DaveS & bob on this


RESOLUTION: 8258 is closed since it has already been addressed by other issues

7728 http://www.w3.org/Bugs/Public/show_bug.cgi?id=7728

Bob: this is proposal6

gpilz: goto section 7.2
... polices appear in meta data section which would apply to messages - what the end point policy - a single policy element can represent metadata second paragraph - Oracle & Microsoft disagree on it

asir: great progress - there are two points misleading part endpoint subject is defined policy attachement (editorial) & second - set the right expectations using the feature - expectations consume is aware or not aware off

<asir> sorry we are in the weeds :-)

gpilz: works in the case consumer knows what the EPR refferes to anybody gives an EPR something can figure it out

asir: anybody is going to disagree

<asir> here is a suggested first sentence ..

<asir> it is desirable for components that provide EPRs to other components that are aware (or can be aware) of service metadata (such as format of messages and transmission protocol) to be able to efficiently communicate the effective policies

DaveS: goes back to my first question - spec that doesnt refer the elements policy that doesnt apply in the constraint - policy into an EPR something can be used it sounds like me something to do the EPR itself it can be stored in secured policy
... concern about putting statement EPR must have an endpoint as subject say that endpoint referes restrict any policy going into it

Tom_Rutt: I am trying to come up with right word - putting a policy in EPR policy attachement with the endpoin the semantic means policy defintion that policy might impact for example whether to use anynomous/non-anynonous end point - approriate do that enpoint policy attachement it is approriate policy menas effecting the smaller levels - attaching policy entire WSDL using it

dug: I think what you are basically say dont pull policy any message exchange with the EPR

<dug> Policy attached to an EPR in this manner MUST be able to be applied to any message exchange using the endpoint referenced by that EPR.

Ashok: do you agree with it

<Tom_Rutt> Any policy in the metadata of EPR has the interpretation that that policy is being attached at the endpoint policy subject level

<Tom_Rutt> It would not be possible to use this mechanism in epr to attach different policy values for different operations supported by that endpoint wsdl.

<gpilz> Policy attached to an EPR in this manner MUST be applicable to any message exchange using the endpoint referenced by that EPR.

<gpilz> http://docs.oasis-open.org/ws-sx/ws-securitypolicy/v1.3/os/ws-securitypolicy-1.3-spec-os.html#_SupportingTokens_Assertion

<Tom_Rutt> You would need to get the whole wsdl for such details, using Mex getWsdl

<Tom_Rutt> The definition of a policy assertion type includes the semantics of what happens when you attach that policy assertion at the entpoint policy subject level. The details of "all messages" "all blue messages" "alll response mesages" etc are defined with the defintion of each policy assertion

Asir: we have to define the policy endpoint subject

<Bob> http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0192.html

<dug> it is desirable for components that provide EPRs to other components that are aware (or can be aware) of service metadata (such as format of messages and transmission protocol) to be able to efficiently communicate the effective policies

<gpilz> Note, for this to be useful the EPR consumer needs to be aware of (or be capable of discovering) additional service metadata such as the format of messages and transmission protocol.

RESOLUTION: 7728 resolved with proposal at http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0192.html


<Bob> http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0188.html

DaveS: It doesnt have creating state
... we have consumer & source

<asir> asir as joined #ws-ra

RESOLUTION: 6436 resolved as documented & pending action on editor to complete client-side table


<Bob> Li's comment http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/0189.html

Bob: empty cells not described by the spec

<asir> this is an appendix

RESOLUTION: 6435 resolved as documented & pending action on editor to complete client-side table

<dug> lhttp://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/att-0193/wsfrag-8196.zip


<Bob> http://lists.w3.org/Archives/Public/public-ws-resource-access/2010Jan/att-0193/wsfrag-8196.zip

dug: scroll down section-6 & one paragraph changed

Bob: Any objection for accepting the proposal

RESOLUTION: close 8196 & apply the resolution of frag as proposed

Rough schedule

Bob: next meeting Feb 9th - All the remaining proposals to be discused & 2119, any questions?
... 2//10 - snapshot , 2/16 - incorporate issues & last call - vote
... 2/23 - published
... 3/2 - Open last call issues

RESOLUTION: Publish first public working draft of WS - Event Descriptons (WS-EVD) specification

<Bob> thank you, Fujitsu for hosting

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/02/09 22:25:20 $