Scribe selection: Paul Lipton of CA (paul.lipton@ca.com)
|
Nr |
Who |
Action Item |
|
3 |
Dave and Lily |
Write proposal for goals and requirements section and send to list |
|
4 |
Steve Graham & David Hull |
Clearly identify WSRF dependencies |
|
5 |
David Snelling |
Submit to WSRF TC to share on the 12th |
|
6 |
William & Peter |
Respond to Martin’s email asserting that TC needs to distinguish between work in progress (editor’s draft) and TC approved (working draft) by email to save TC meeting time. |
|
7 |
Peter |
Send list of typos in WS-Topics to William |
|
8 |
Igor |
Write up proposed NotificationProducer clarification |
|
9 |
Sanjay |
Write up the issue of Simultaneity and the semantics of setting subscription properties (first review with Dave H.) |
Please note:
If you join late and miss the roll call, please email the chairs ASAP to be
added to the attendance list. This will
avoid interruptions during the meeting.
Peter: Call
was on the 7th, minutes say 2nd.
William:
filename also has error.
Peter: I will
update and mark the changes.
Peter: Look at
home page (TC policy). Also look at voting. IMO, too much latitude could cause
problems. The rules are attend 2 out of 3 meetings
within any rolling window of 3 meetings. Failure will result in warning to
require attendance on next call. There will be a second warning if you
subsequently fail to show (relaxation of rules). If you fail to show after the
second warning you will be reset to observer status. You have to vote in 2 out
of every 3 ballots. You can always abstain, so it’s easy.
Dave Orchard asked
a question about national holidays.
Peter:
Generally we will not schedule calls on US holidays. National holidays in other
countries won’t count against you, if you send in notice to chairs
explaining the circumstance (preferably ahead of time). Leaves of absence open
to all (1 per 12 months). Apply to chairs for this privilege. Check OASIS
website for details.
Jeff M: Are
you going to distinguish between informational and technical (binding) votes?
William: I
agree we should distinguish non-binding (will be clearly stated). The default
is mandatory (it counts).
Jeff M: What
about F2Fs?
Discussion
about how these should be counted (one meeting for whole F2F or one meeting for
the n number of days of the F2F). Consensus is to be F2F as one meeting.
Question about
dial-in: are people expected to dial in each day? The suggestion that we keep
the line open all the time, but provide a focused time slot for votes and
discussion of critical issues was generally agreed upon.
Peter: No
meeting of WSN that day as per above policy.
Steve Graham:
How about sharing with WS-RF?
David Hall:
How about Tuesday?
William:
Better to share with WS-RF.
Peter: We will
have a gap of 3 weeks. Agree on July 12th (1 hour after WS-RF meets at
William: We
should add Igor to list. [SCRIBE NOTE: Is this an action item?]
William:
Worried about introducing new editor’s draft. Agree with Martin that it
would be nice to avoid small incremental versions. Open to considering ways
that are not too heavy (special folder?). Don’t see benefit of a new
class of documents.
Martin: Need
to distinguish between work in progress (editor’s draft) and once
approved (something like working draft).
William: We
don’t expect many more votes.
Martin:
Disagree. Need periodic clarifications and votes. Editor’s work is
editor’s work, not the committee’s work. The only way to determine
that the editor’s work is good is to take a vote. The committee needs to
take a vote, periodically.
Peter: I
thought the issues list takes care of this.
Martin: The
vote is a close off and validation.
William: But
why a special status?
Martin:
It’s a really simple way to say “that’s the latest
baseline.”
Peter: We do
have set of stable working drafts that are distinguished from what led up to
them. We’ll do a separate folder. Let’s see how this works.
Martin: I
object. We won’t know the version when we talk. WS-I and W3C and others
use editor’s draft (unapproved) and working draft (committee’s
approved draft).
Jeff M:
Editor’s draft not approved by TC yet.
Fred:
We’re arguing about documents we don’t have.
Jeff: When
somebody makes an edit, who owns the document?
William: So
you want a mention on the document when it has been voted on?
Jeff: Until
the group votes upon it, the editor’s documents have no standing.
William: They
want to see on the download what the status of the document is.
Peter: An
issue isn’t the same as a vote on the actual correction made from that
issue.
Martin: But,
there has to be a way to indicate baselines that the group is happy with.
William:
Worried about doing this all the time will slow down the group.
Peter: What
about just opening another issue if you don’t like how an issue is
resolved?
Jeff: The TC
has to have a proper check on appropriate editorial discretion.
Brief
discussion about naming of drafts. How should we distinguish between
stable working drafts and work-in-progress drafts?
- WS-BaseNotification ballot result
Peter: Ready
to go, but additional acknowledgements need to be added. OK to go without
additional vote?
William: No
opposition “this document was approved on (the date the ballot
closed)” being added to document. Text of ballot indicated this sort of
minor change might be added.
No opposition voiced to this by attendees.
Lily: Wrong
reference listed. Don’t we have to put in the requirements stuff?
Peter: Yes. Is
everybody happy with requirements?
William: When
requirements done?
Dave and Lily:
Friday.
- WS-BrokeredNotification
draft
- WS-Topics draft
Peter: found
3-4 typographical errors that will be posted. Original doc had bulleted lists
that have been moved.
William: I did
copy/text and may have missed some.
Agreed
that with those typo fixes we could go to vote.
NotificationProducer properties [6]
Peter: Raised by Igor. Email of 14th.
Igor: Simple
error. Fragment in spec says how the property should look. XML can’t look
like that. Schema is correct. But, wording is wrong. I suggest rewording of
spec.
Steve Graham:
Agree it is misleading.
Igor: These
elements described in fragment don’t exist anywhere (in schema or
anywhere else). Either use the fragment (fix) I sent
or have an instance of how the property document will look.
William:
Igor’s right. It’s not really correct. Just put a ref directly to
the topics.
Igor:
Currently, as exists, there is no way to use it. Those element names exist, but
the cardinality does not. Just to make it coherent with current schema, make
some change. Having an actual schema (rather than a pseudo-schema) is very
helpful.
William: Your
proposal, if we just show instance, people could use schema definition to find
out what is needed instead of helpfulness of cutting/pasting right from spec (a
convenience).
Steve Graham:
I prefer Igor’s approach, you don’t get
much with pseudo-schema.
Peter: But,
with Igor’s approach you can’t see the datatypes.
Right?
Igor: I think
reusable is better.
Steve Graham:
Could have two snippets to solve this. Then you doc
cardinality, too.
Allow NotificationProducers
to refuse subscription requests [7]
Peter: We want
not always accept requests (security and so on). Already
added as number.
Documenting schema pseudo language [8]
Background
from Steve: We’ve never described the pseudo-schema we use.
Peter: Steve
was going to look into this.
Steve:
Resolution was posted. This issue is problem in other specs like WSDL 2.0
working group. He discussed with other editors. We should use approach of WSDL
WG.
William:
They’re hoping to do this before August.
Jeff:
I’m a little skeptical.
Steve: If we
can’t wait, we’ll craft our own language, but prefer not to.
Peter: Goes in
chapter one, probably. Just one place, not in all docs.
We’ll wait to see what we get.
Formally added
to issue list now is suggested.
Simultaneity and the semantics of setting
subscription properties [9]
David Hull:
Main issue is that we allow change of subscription destination on the fly.
Probably not problems with changing topic set and other things, but when we
allow destination changes that is a challenge in terms about atomicity. Might be problem for some implementations (much more weight).
Proposal to say destination cannot be set as property, but other things are OK.
We need to be clear about semantics.
Sanjay: If
subscriber was aware that there are multiple
broker…
David Hull:
Subscriber shouldn’t be aware.
Sanjay: What
if request needed to be sent to two different subscription managers. Simpler
model is subscriber just knows about one subscription manager. Weight is
heavier, but spec is simpler.
DH: When you
cancel a subscription do you talk to a different thing than when you made it is important, but a different issue.
David Hull:
Atomicity is not really the best term, of course. Each message sent arrives
exactly once someplace. We shouldn’t require this for all
implementations.
David Hull: We’re
pushing more and more to policy, which I agree with as long as we don’t
forget about it. We don’t want to be normative about this, but we may
want to give advice.
Steve G: We
could be normative on how notification producer advertises.
David H: There
are only a few things that make sense.
Peter: If we
could get some scoping, that would be great.
DH: Well, one
is distinction between producer and subscription manager (Sanjay?).
Sanjay: What I
said is different. MISSED A LITTLE OF WHAT SANJAY SAID.
David H: Need
to clarify what the implementation required to do and what the semantics are.
Generalise WSBN to cover indirect notification
patterns [10]
Peter: See
wording in intro lines 105 to 110.
David H:
Already generalized. Presents same interface. Wording
in introduction is non-normative.
This will be revisited on meeting of 12th.
REFERENCES:
[6] http://www.oasis-open.org/archives/wsn/200406/msg00050.html
[7] http://www.oasis-open.org/archives/wsn/200406/msg00059.html
[8] http://www.oasis-open.org/archives/wsn/200406/msg00054.html
[9] http://www.oasis-open.org/archives/wsn/200406/msg00067.html
[10] http://www.oasis-open.org/archives/wsn/200406/msg00065.html
[11] http://www.oasis-open.org/apps/org/workgroup/wsn/download.php/7258/WSN_IssuesList1.18.doc