Minutes of WSA informal telcon on 2002-08-01

BEA Systems David Orchard
ChevronTexaco Roger Cutler
France Telecom Shishir Garg
Idokorro Mobile Mark Baker
IONA Eric Newcomer
MartSoft Corp. Jin Yu
Nortel Networks Abbie Barbir
SeeBeyond Technology Corp Ugo Corda
Sun Microsystems, Inc. Doug Bunting
W. W. Grainger, Inc. Daniel Austin
W3C Hugo Haas
W3C David Booth
Daniel is scribe
Mike C posted an informal agenda
Mike C: requirements document has been updated, changes by Chris F, Sharad, Daniel
Mike C: can't yell at CHris, he's not here
Mike C: we don't intend to make an substantive changes in AUgust for summer holiday
Mike C: we are allowed however to make editorial and administrative changes
Mike C: some of CHris's changes were incorrect
Daniel isn't typing
Zakim, who's talking?
Daniel: I saw there was confusion about changes that were made.
dougb, listening for 10 seconds I heard sound from the following: +1.603.863.aaaa (13%), Hugo_Haas (14%), ??P7 (100%), ??P8 (85%), DOrchard (4%)
... I'll volunteer to make the corrections to it.
Mike C: we need to get this out in hours for Sept. publish?
Hugo: no, it's a working draft, let's not delay for a few small changes
Hugo: asks if Daniel cannot live w/ doc as it is
Daniel: no, I was just volunterring to fix it if it was an issue
Mike C: I'm happy with it, unless there is a dealbreaker, let's leave it
Daniel: ok
Mike C: does anyone else have any substantive problems with it, as it is?
Daniel: we will publish early Sept. wth changes
Hugo: that's fine
Mike C: hearing no diagreement, let's go ahead
Mike C: Do we need to modify dates?
Daniel: Hugo, if you tell me the pub date, then I can fix the date. I was going to fix HTML errors to make it valid.
Hugo: Make it next Wednesday.
... Running Tidy will prob fix the HTML.
Daniel: I have a stylesheet that fixes the HTML also, and I'll post it to the group.
Daniel: document date will be AUg 7
Hugo: also change previopus versions URLs
Daniel: ok
ACTION: Daniel to make mods to reqs document (what else is new?)
Daniel, type "ACTION: Daniel to ..."
thanks :)
Mike C: end of reqs doc issues
Mike C: we have very few ppl here today, we should not make substantive decision til sept.
Daniel: do we have a quorum?
Mike C: I don't think so...we are just talking
Mike C: Chris says we have a moral requirement to publish the arch doc asap
Daniel: if we have a quorum, why can't we make decisions?
Daniel: Why are we unable to make decisions in August if we have a quorum?
Mike C: we agreed to that a few weeks ago
Mike C: understanding that we are on break in August
Mike C: lots of vacations in August, so no official decisions
Mike C: we want to start crystallizing consensus on arch doc w/o making decisions
(scribe notes that this is sketchy)
Mike C: we currently have almost no idea what to do with the arch doc
Mike C: I suggested some options in my recent email
Mike C: gather references, as examples
Mike C: we sent mail to the OMG
Daniel: is this a task for the harvesting team only?
Mike C: general guidance for editors on how to articulate a reference architecture
Mike C: sent pointers to MS also
Mike C: what other things can we use for references?
Mike C: have members written other white papers etc?
Daniel: IBM has some papers
this? http://www.research.ibm.com/journal/sj/412/gottschalk.pdf
Mike C: there is also some material from last year's workshop
Mike C: we need to ratchet up the level of the discussion of the document
MikeC: and move away from the abstractions we have been discussing
Mike C: I asked DaveO re BEA's WS products
Mike C: Dave, do you have a URL?
Mike C: anybody else?
IONA: we submitted a paper to the workshop
Mike C: that's a good thought
IONA: I think that's why the WS activity was started originally
Mike C: we should perhaps read these again
IONA: we also talked about future directions, perhaps that would help
Mike C: Mark B is there anything in Roy F's miserable thesis that deals with this?
Mark B: I think I would just go with the thesis
scribe notes that :"miserable", while a worthy description, was not part of Mike C's speech
Mike C: do we think this is a useful exercise?
Mike C: we need to understand each other's views to achieve consensus
D ave B: I think it's useful, but I wonder if it will take too long
starting point: http://dev2dev.bea.com/techtrack/detail.jsp?highlight=webservices
Mike C: speaking of harvesting, I am still catching up, what is the status? we've heard about REST but that's it
Mike C: I haven't heard from many of the other things we have discussed
Mike C: can anyone give us an update?
Dave B: there's been a lot of work, ebXML, WSDL, REST
Dave B: I don't think we are ready to move forward, maybe DaveO's suggestion to move forward with SOAP is a good idea
s/Dave B/Mark B/
Mike C: it might be helpful...when was it posted to the mailing list?
s/Dave B/Mark B/
Should point out that many of the papers at http://www.w3.org/2001/03/WSWS-popa/ aren't architectural though they may be descriptive.
LOL scribe blows it, thinks Dave and Mark are interchangable
s/Dave B/Mark B/
NOte: above references to Dave B are to Mark B, apologies from scribe to both
Mike C: when was your message posted Mark?
I'd recommend having us (the members) pointing towards specific papers rather than asking all to read every WS Workshop paper.
ACTION: Mark B to repost his harvesting discussion
Mike C: some of the people involved in harvesting besides Mark, any thoughts?
Mike C: what should the results of the harvesting be? links? pictures?
Mike C: summary papers? ideas from others?
Group: silence
Mark B: there are two major features that we harvested
Mark B: we also harvested or tried to harvest architectural elements, and got quite a bit done
Mark B: we should write this all done
Mark B: the current document identifies style elements, but not architectural elements
Mark B: I've proposed a new section to the doc to remedy this
Hao: should we use standards here?
[ I agree with Mark but won't try to unmute... ]
Mike C: mo ichido kudasai?
Hao: what notation should we use?
Mike C: UML comes to mind, but I'm not sure...
+Daniel dammit
Zakim, add me to the bloody queue you foolish piece of trash
I don't understand 'add me to the bloody queue you foolish piece of trash', Daniel. Try /msg Zakim help
Dave O: UML was brought up in the TAG< and rejected, for good reasons
Dave O clarifies: wasn't rejected, was some pushback but no decision was made
Mark B: Dave suggested Krutchen's 4+1 view but Roy F. pushed back
ACTION: Daniel to post 4+1 model
fwiw, Roy's view on 4+1; http://lists.w3.org/Archives/Public/www-tag/2001Dec/0023.html
Daniel: Someone earlier suggested that we used UML, and I pushed back because I don't think UML captures architectural models well. I'm happy with a simple box model. I think we can establish all the ideas we need without any particular choice of notation.
Daniel doesn't want to put "don't use a notation" as an intersectional approach in the notes. Sorry Daniel...
Hao: I was just thinking that we can just identify some use cases, independent of notation, and begin
LOL Daniel submits the idea that we should not choose any notation
Hao: start with something very simple, and work through use cases, avoiding notation wars
Question for the Harvesting team: Are we revisiting a discussion that's already happened within that group?
DaveO: Mark correctly points out that we have only a functional approach in the current document
DaveO: Mark and I may disagree on certain elements in the architecture
DaveO: we need to deal with two issues: a) how are SOAP, WSDL extended and b) how they are used by ppl
DaveO: there is a very specific model in WSDL, that we can talk about
zakim, who is barking?
I don't understand your question, MarkB.
DaveO: in the security spec, there are specific extensibility mechanisms, but no real way of specifying the means by which extensions are implemented
DaveO: there is no processing model, so there is much duplication of the SOAP processing model
DaveO: suggests principles of extension would be very useful in the arch doc
Mike C: I initiated (with Mark on the mailing list today) trying to find the intersection fwhat we can all agree upon
Mike C: such as accepting the SOAP processing model, which may lie in that intersection
Mike C: how can we make this happen?
Mike C: do we need additional discussion? how do we move ahead?
DaveO: the way I think we should do this is similar to the way the TAG has done, specfying protocols, etc.
DaveO: the principles are most important, and establishing them
DaveO: e.g. using URIs in SOAP
DaveO: may books have verbiage about use of URIs, including character level information and cannonicalization
Daveo: this might go into the identifiers section
DaveO: also combining specs that other ppl have created
Joe Hui: there is still a potential conflict between SOAP processing model and REST model
DaveO: REST doesn't have a processing model at the same level of abstraction as SOAP
Joe Hui: aren't orchestration and processing model related?
Doug: I think that what we are disucssing is the layered model
Doug: REST by design doesn't talk much about processing behind the HTTP server system boundary
Doug: SOAP explicitly does go there
Doug: it may be that we need to include the processing model in our discussions
Doug: we will have to extend the REST model
Mark B: responds to Doug - as Doug say, SOAP extends REST, because of the circumstances, no explicit processing model in REST
Mark B: SOAP is much richer than REST but not radically different
Mike C: Soap is more about server to server communication
Mike C: other models include the back-end processing
Mark B: SOAP by itself doesn't buy you very much, it's only when it is used by an application
Mike C: I was thinking of CORBA or DCOM
DaveO: why can't we talk about SOAP when going from server to server?
zakim, who is here?
On the phone I see ??P0, ??P1, ??P2, ??P4, ??P7, MarkB, Dbooth, +1.650.875.aacc, +1.408.737.aadd, DOrchard, ??P14, Hugo_Haas (muted), ??P15, ??P8, Joseph_Hui, Tim_Jones
On IRC I see Tim_Jones, Ugo, dougb, DaveO, mnot, dbooth, Daniel, RRSAgent, Zakim, MarkB, Roger, mchampion, HFN_away, hugo
Mike C: I didn't mean that...this is only one aspect of the system, others such as security, orchestration, etc. are also important to what we are doing
Mark B: I think the whole server to X concept is a red herring
Doug apologizes for raising a red (or pickled) herring.
Mike C: Mark it sounds like you are more or less comfotable with Dave's doc at this point...
Mike C: picking up from there, one suggestion to consider would be to come up with a written record of what you think the processing model might be
Eric: I could try to help
Dave O: that is what I had started to do, I introduced the notion of two agents communicating, usually with XML, and then
Dave O: we started talking about the interactions and the formats and protocols to be used, this is a typical means of beginning
Eric: the SOAP model has some of this material, but not all
Dave o: what are the appropriate things to take out of SOAP and WSDL and then stitch them together
*scribe thinks this is a frenkenservice*
Eric: not sure to what extent this might be normative, but we should have that discussion
Mike C: we need to find some common ground on which we can all agree
Mike C: I agree with what Dave O said
DaveO agrees with puttin' pen to paper
Mike C: does WSDL have a processing model?
Eric: no
Dave O: no, and there is not much to tell people how to use it
Eric: WSDL doesn't even talk about SOAP
Hugo: yes it does
Dave O: we need to find a way to relate the processing model to WSDL
NB: was mnot, not hugo
Eric: a lot of this is informative material rather than normative
(I think hugo is permanently muted? ;)
scribe apologiezes to mnot
no worries
Mike C: this is turning into an action item to the group to consider this...how should we proceed Dave, Eric
Eric: I'm happy to do so
Mike C: does this remind anyone of any other white papers or other sources of into
Eric: I wrote a book on WS, I can look through it
Dave O: I encourage ppl to submit info
Doug: the workshop materials are useful
DaveO: thinking has changed since then also
Eric: there is some data from MS, and others
Dave O: perhaps the chair should ask members for materials
ACTION: Mike C to query members w/regard to submitting their own materials on WS-arch
Mike C: Mark B are you comfortable with this?
Mark B: I think so
Eric: Mark please help us with this
zakim, who is talking?
Mark B: okay
zakim, who is talking?
MarkB, listening for 10 seconds I heard sound from the following: ??P1 (10%), DOrchard (14%)
scribe unable to understand the call, due to noise
dbooth, listening for 10 seconds I heard sound from the following: ??P1 (39%), ??P4 (39%), MarkB (39%), DOrchard (9%)
MarkB: i'll be glad to help, and don't want to spend too much time on REST
.. for now 8-)
Mike C: I think it's useful to get a document published
Mike C: many press articles are not concentrating on these issues
Mike C: first step is harvesting
Zakim can't think as fast as I can type Dave
Mike C: we need to make progress on principles, esp. URIs
Mike C: what is a good way forward?
Notes it was Mike C who thought the workshop materials were generally useful and Doug who suggested reducing volume thorugh "submissions" process (some lines back).
Mike C: does anyone diagree?
mark B: we should consider both long and short term views
? URI's still used improperly in SOAP 1.2 even after GET addition?
Mark B: URIs are certainly a serious issue, but we need to decide what the short term goal is
thanks DOug
Mike C: let;s try to identify the low hanging fruit for the short term
Mike C: URIs may be an issue that we don't have to do lots of work on
Mark B: there are several features of URIs but we should consider what's important
Mark B: I need to go, sorry
Dave O: One of the things I want the document to take into account is what we've all learned rrecently
Dave O: we should be agressive about items like URI usage etc.
Dave O: e.g. we should provide some guidance about how to use these things
Dave O: like URIs
Mike C: what does that mean operationally?
Mike C: do we need to include previous arch discussions? Agressive how?
Dave O: we should write these things up explicitly to provide guidance to users
Dave O: I thin of WS-ARCH as an extension of existing technologies
Dave O: we should be clear about how to use these things, and explicit, e.g. stating which protocols to be used
Mike C: anyone else have any perspective on this?
Hao: we have done some harvesting for WSDL...are we going to do anything about the results of this?
Mike C: asks to repeat
I particularly like Dave O's comments about both refinements and extensions (of REST in particular) inherent in our proposed web services architecture.
Hao: we've harvested a lot of stuff, what do we do with it?
Mike C: my understanding is that the harvesting is more of a collective activity
Mike C: and the editing is the domain of the editors
Daniel needs to leave...can someone else scribe for the remainder of the call?
I can make an attempt to fill Daniel's shoes.
Thanks Doug!
DaveO: There is no try, only do...
see you all next week
YodaO: no try, only do!
MikeC: Harvesting source for how we'll evaluate the editor's output (our architecture)
Hoa: Wants something more concrete.
zakim, who is talking?
MikeC: What are u suggesting.
dbooth, listening for 10 seconds I heard sound from the following: ??P1 (19%)
zakim, please mute P1 temporarily
sorry, dbooth, I do not see a party named 'P1'
Admonition from the chair (MikeC): Comment for the record: Do not put yourself on hold when you must step away if you have muzak on hold.
Hao: Make it more official, include harvesting material as an appendix.
MikeC: you're suggesting we pull this stuff together into an actual document
Hugo: I can be heard!
Hugo: Mainly harvesting material should be used as a checklist: See what we forgot (which concepts) and how they might fit.
MikeC: Might mean we have to worry less about appearance of this material.
MikeC: Comments on this? Any other words of advise?
Zulah: Did we talk about requirements document?
MikeC: That document will be published pretty much as is but with eye towards another publication early in Sep
Zulah: Objects to this publication (the one next week) due to missing decisions (agreements and approvals not reflected in the document).
MikeC: Chris F. and Daniel not on the call. Concensus was things would be better in an ideal world. This was just a working draft.
MikeC: May even publish again first week we're back in Sep
Zulah: Are minutes the official record?
MikeC: They are but editors sometimes miss a few agreements. he'll track these issues and work with editors to address in later (Sep) version.
MikeC: No problem with publishing requirements as (another) working draft?
Zulah: Not the first working draft (second).
Hugo: No objections to Zulah forwarding comments to Daniel and / or Chris. Don't hold publication because of this. "Just" a working draft.
MikeC: If Daniel can make changes without changing schedule, great. If he can't, also fine.
Hugo: Basically, publish or perish.
Zulah: Will send email to editors. Making point we're working towards a deadline (eg. RTF deadline was met but materials didn't go into document).
MikeC: Issues may have to do with transition of chairs. He was pushing harder on some things and less on others. Will improve process going forward.
Zulah: Comfortable with the decisions regarding comments, changes and this publication.
**** Call ends ****
Zakim, please excuse us
RRSAgent, please excuse us
I see 4 open action items:
ACTION: Daniel to make mods to reqs document (what else is new?) [1]
recorded in http://www.w3.org/2002/08/01-ws-arch-irc#T19-51-12
ACTION: Mark B to repost his harvesting discussion [2]
recorded in http://www.w3.org/2002/08/01-ws-arch-irc#T20-09-17
ACTION: Daniel to post 4+1 model [3]
recorded in http://www.w3.org/2002/08/01-ws-arch-irc#T20-16-36
ACTION: Mike C to query members w/regard to submitting their own materials on WS-arch [4]
recorded in http://www.w3.org/2002/08/01-ws-arch-irc#T20-43-23