Meetings/Telecon2014.05.29

Event details

Agenda

Minutes

Chair
Arnaud J Le Hors (IBM)
Scribe
John Arwe (IBM)
Attendees
Arnaud J Le Hors (IBM), David Green (Tasktop Technologies), Harish Krishnaswamy (Software AG), Jim Conallen (IBM), John Arwe (IBM), Nick Crossley (IBM), Sam Padgett (IBM), Steve Speicher (IBM), Martin Pain (IBM)
Resolutions
  • Minutes of 15 May approved unanimously
  • Next TC call will be on 12 June 2014
Actions
  • Nick Crossley to send a proposal to define a context HTTP header

Original Chat transcript from room: oslc

2014-05-29 GMT-08:00 (Timestamps below are US Pacific)

[07:04] Nick Crossley (IBM): On Skyoe at an airport today, so will be on mute most of the time. Need to leave in about 20 minutes. Would like to have a quick discussion on the UI preview width resizing if possible,

[07:04] Nick Crossley (IBM): Skype, that is.

[07:06] John Arwe (IBM): wrt that Nick, I was not even aware of the asymmetry. I assume Core 2 is merely silent (vs MUST NOT) on one aspect. I'd certainly prefer them to be symmetric, all other things being equal.

Roll Call

[07:08] Arnaud J Le Hors (IBM): roll call: Arnaud J Le Hors (IBM) David Green (Tasktop Technologies) Harish K (Software AG) jim conallen (IBM) John Arwe (IBM) Nick Crossley (IBM) Sam Padgett (IBM) Steve Speicher (IBM) Martin Pain (IBM)

Approval of May 15 minutes

[07:08] John Arwe (IBM): Speicher looked, updated some formatting.

[07:08] John Arwe (IBM): Arnaud assumed those updates were gremlines

[07:09] John Arwe (IBM): Unanimously approved

Next meeting

[07:09] John Arwe (IBM): next call, +2 weeks, June 12

[07:09] John Arwe (IBM): No conflicts asserted

Update on LDP

[07:10] John Arwe (IBM): WG finished work on main spec for now, is asking for transition to Candidate Rec, exit criteria = 2 impls of each feature.

[07:12] John Arwe (IBM): Test suite WIP, on github. CR draft not yet published; trying to schedule Director's call as required. No formal objections expected, one slightly unhappy commenter. Assume CR draft will emerge w/in next 2 weeks.

[07:13] John Arwe (IBM): No min/max for "in CR" state; lots of impls in progress, so could be on the short side (w/in another month or 2); as soon as we have impls, exit CR and then PR with rep voting

[07:13] John Arwe (IBM): Paging was split off, on separate timeline.

[07:15] John Arwe (IBM): Not sure how long paging will take; depends on WG focus/consensus.

[07:15] John Arwe (IBM): Speicher: also separate Patch format being worked in parallel.

[07:17] John Arwe (IBM): LDP encourages patch support, but specifies no format (and no other std one exists). Taking last-gasp effort to define one on a "more or less LDP 1.0" schedule, rather than waiting for LDP+1 which could be >1 year.

[07:17] Steve Speicher (IBM): Other interesting LDP impl stuff going on (yet to do much eval of it) https://github.com/ldp4j/ldp4j and morph-LDP http://2014.eswc-conferences.org/sites/default/files/eswc2014pd_submission_85.pdf

[07:17] John Arwe (IBM): Sub-group said "nearly done" a few weeks ago, then went underground while they try and make it more complicated ;-)

UI resizing

[07:19] John Arwe (IBM): Nick: anyone know why 2.0 missed it? 2.0 silent on one for UI preview, explicitly allows both for pickers.

[07:19] John Arwe (IBM): Speicher: don't know; might have been simple absence of use cases. Agree same resizing model should apply to both dialogs.

[07:20] John Arwe (IBM): ... agree to add/generalize it.

[07:20] Nick Crossley (IBM): Need to leave now to board my flight. For width resize, would like Core 3.0 to be symmetrical between pickers and UI preview. For configuration context, I'm quite willing to do the work for that (such as an RFC if necessary), but Im also OK with using a Link header instead if the group prefers that.

[07:21] John Arwe (IBM): Nick: did see this morning's email from Arwe on Link/Vary

[07:21] John Arwe (IBM): Arnaud: for resizing, who's doing that?

[07:21] John Arwe (IBM): Sam: I'm already in that spec, will handle it.

[07:22] John Arwe (IBM): ... names do not conflict, so should be possible for one impl to support both if it wants to be compatible with both sets of clients.

[07:22] John Arwe (IBM): Nick leaves for plane

Namespaces

[07:26] John Arwe (IBM): Arnaud: person we need for this is unavailable - accident. OASIS thinks namespaces should change, so far. To some degree we had observed apparently conflicting guidance "this is the norm, but there are exceptions". Working through what the effective policy really is, what Should qualify as an exception.

[07:27] John Arwe (IBM): Speicher: PROMCODE WG were the ones asking; chair not here, located in Asia so Core meets at bad time; should my response for now be to follow the guidance on our wiki page, and then we'll adjust if needed once the OASIS discussion can restart (and conclude)?

[07:27] John Arwe (IBM): Arnaud: yes

Delegated UIs

[07:30] John Arwe (IBM): https://wiki.oasis-open.org/oslc-core/DelegatedUIUseCases#preview

[07:31] John Arwe (IBM): Speicher: IBM CSI area came up with "theme name" idea, passed into dialogs/previews. Any others?

[07:31] John Arwe (IBM): [crickets]

[07:33] John Arwe (IBM): ... if no one else has hit this, maybe no clear need to standardize it. Within Rational we have a common style/theme, but as we built adapters the question was which of the two "sides" was the adapter closer to? Tended to end up looking most like the data source, and manually toning down any obvious conflicts.

[07:33] John Arwe (IBM): Sam: philosophical and design aspects

[07:34] John Arwe (IBM): Speicher: CRYSTAL project from PTC brought this up, they tend to try integrating contrasting products

[07:34] John Arwe (IBM): ...option: send msg to dialog telling it what to use; unnatural ... load it then update it

[07:35] Harish K (Software AG): I got bumped off the call - rejoining in a bit

[07:35] John Arwe (IBM): ...option: input parameter to dialog; more natural, allows server to change content before first rendering if needed

[07:36] John Arwe (IBM): Arnaud: trying to find middle ground; not sure it's worth doing, hard to find middle ground.

[07:36] John Arwe (IBM): Martin: we also made adapters look like "the other tool" so users are conscious where data lives

[07:37] John Arwe (IBM): Speicher: Arnaud seemed to be saying something different

[07:37] John Arwe (IBM): Martin: we expect the UI to look like the data's server

[07:38] John Arwe (IBM): Speicher: only ctl I'm talking about it is whether the page embedding the iframe gives the server a hint (like CSI did), and the server is free to use it or ignore it

[07:38] John Arwe (IBM): ... input could be a "theme" name or a css file name, for example

[07:39] John Arwe (IBM): Arnaud: the css case is what I'm worried about - don't want client coding css, encourages tight coupling that's contrary to the point of delegated dialogs

[07:40] John Arwe (IBM): Speicher: no, server still does rendering. the css option doesn't mean the client has to code anything.

[07:40] John Arwe (IBM): ... loose control over styling.

[07:40] John Arwe (IBM): Arwe: hint to server, ala Accept header?

[07:42] John Arwe (IBM): Speicher: yes. 1 issue I have is we don't have much experience with the solution. with css could get into lots of trouble.

[07:42] John Arwe (IBM): Arnaud: +1

[07:42] John Arwe (IBM): Speicher: "theme" idea suggests that server should communicate list of themes it supports to the client

[07:43] John Arwe (IBM): Arnaud: that seems pretty safe. if server is given style sheet, you have no idea of the impact.

[07:43] John Arwe (IBM): Arwe: clickjacking risk => servers unlikely to allow

[07:45] John Arwe (IBM): Arnaud: seems like too early to put something in the spec anyway; experimentation needed.

[07:45] John Arwe (IBM): Speicher: that's why was hoping someone had great solution sitting on the shelf already

Config context header

[07:47] John Arwe (IBM): Arnaud: summarizes history

[07:49] John Arwe (IBM): Arwe: summarizes his email from today

[07:49] John Arwe (IBM): Arnaud: no Nick, no Ian, just Nick's parting comment above

[07:51] John Arwe (IBM): Speicher: Vary says that caches have to treat requests with different Links as different requests, and separate the responses. So we're really pushing on how many Link headers we expect on future requests.

[07:51] John Arwe (IBM): Arnaud: how do we make progress? strong feelings?

[07:52] John Arwe (IBM): Speicher: separate header = clean + clear, but more process to go through IETF; since Nick's willing to go the IETF route, "why not?"

[07:53] John Arwe (IBM): Arwe: my only "fear" is that IETF will think we're crazed, but why not let that play out? if they're ok with it, we're done.

[07:54] John Arwe (IBM): Arnaud: one adv of IETF is the wider review; either they poke a real hole in it, or we get the new header and done

[07:54] John Arwe (IBM): Consensus on NOT prefixing the new header with OSLC-

[07:55] John Arwe (IBM): ...and taking it to IETF

[07:55] John Arwe (IBM): Arnaud: Nick to write up formal proposal incorporating that discussion for next call, and we can vote on it

[07:56] John Arwe (IBM): ... in parallel if he wants he can start the IETF wheels turning

[07:58] John Arwe (IBM): no objections heard

Other business

[07:59] John Arwe (IBM): Jim C: are transactions a queued topic anywhere? we're re-looking at how we're doing it in my product right now. is this the forum to do it? or somewhere else?

[07:59] John Arwe (IBM): Speicher: it has come up in LDP; we can talk about it here

[08:00] John Arwe (IBM): Arnaud: not sure what you mean by trxn in this context; "http = stateless" response tends to come up

[08:00] Martin Pain (IBM): I need to drop off

[08:01] John Arwe (IBM): JimC: we have 10s-100s of thousands of interconnected resources that needs to be consistent; can take 6-8 hours on a bulk load before everything is consistent.

[08:01] John Arwe (IBM): Speicher: maybe write up use case

[08:01] John Arwe (IBM): JimC: ok

[08:01] John Arwe (IBM):

[08:01] John Arwe (IBM): Arnaud: anything else?

Adjourned

Meetings/Telecon2014.05.29 (last edited 2014-05-29 20:43:20 by alehors)