Draft Minutes 4th September 2013

[Adrian Otto <adrian.otto@rackspace.com>]

Meeting Minutes 4th September 2013

Attendees:

US Department of Defense (DoD) Michael Behrens  Voting Member
Software AG, Inc.         Bhaskar Reddy    Member
Oracle                         Mark Carlson     Voting Member
Fujitsu Limited                 Jacques Durand   Voting Member
Oracle                          Anish Karmarkar  Voting Member
Oracle                          Ashok Malhotra   Voting Member
Rackspace Hosting, Inc.         Adrian Otto      Pro-Tem Chair
Vnomic                          Derek Palma      Voting Member
Oracle                          Gilbert Pilz     Voting Member
Red Hat                         Krishna Raman    Voting Member
Fujitsu Limited                 Tom Rutt         Voting Member
Software AG, Inc.               Prasad Yendluri  Voting Member

Voting Members: 11 of 12 (91%)

Achieved quorum: yes

Intro: 

Martin Chapman is having an LOA, as announced in https://www.oasis-open.org/apps/org/workgroup/camp/email/archives/201308/msg00056.html

No objections to proposed agenda. 

Agenda is accepted.

Prior Minutes:

MOTION: m:Gil, s: Mike: move to approve the minues from 8/28/2013 https://www.oasis-open.org/apps/org/workgroup/camp/email/archives/201308/msg00091.html

Motion passes by UC

TOPIC: Plugfest (Both Mike and Gil will attend in Santa Clara)

Gil Pilz (Oracle): http://www.cloudplugfest.org/cloud-interoperability-week
Gil Pilz (Oracle): Mike: is there a CAMP server impl?
Gil Pilz (Oracle): Gil: nCAMP is up

Gil Pilz (Oracle): nCAMP is at: http://ec2-107-20-16-71.compute-1.amazonaws.com/campSrv/

TOPIC: Issue Discussion

TOPIC: CAMP uses undocumented notation to describe resources http://tools.oasis-open.org/issues/browse/CAMP-51 (Proposal In JIRA)

Gil Pilz (Oracle): Anish: there were conditions in which everything captured in the pseudo-schema was not captured in text
Gil Pilz (Oracle): ... is this still the case, and what should we do if so?
Gil Pilz (Oracle): ... in addition to ?, *, and + - we don't have a symbol for "exactly one" - maybe that needs to be explained
Gil Pilz (Oracle): Gil: first issue should be fixed - we should say in text everything in pseudo schema
Gil Pilz (Oracle): ... exactly one is sort of obvious
Gil Pilz (Oracle): Anish: we should add a bullet point in the notional convention that explains "exactly one"
Gil Pilz (Oracle): Gil: you and I can work offline and figure out what that sentence should look like
Gil Pilz (Oracle): Jacques: the scoping of operators (?, *, +) is funny - we don't use parentheses
Gil Pilz (Oracle): ... maybe we should say something about this?
Gil Pilz (Oracle): Gil: should we just strike the bullet point about parentheses?
Gil Pilz (Oracle): Jacques: it might be useful to be a bit cripser - "the default scope is the entire line"
Gil Pilz (Oracle): ... the "|" symbol has a more limited scope
Anish Karmarkar: (Tom is sitting next to me and this is his suggestion characters are appended to items to indicate cardinality (absence of an appended character indicates the item is present exactly once)
Gil Pilz (Oracle): ... it is used only once
Gil Pilz (Oracle): ... in Section 4.6
Gil Pilz (Oracle): Gil: brackets in 4.3.6 are a bug - it's not an array
Adrian Otto (Rackspace): Link to Gil's proposal for CAMP-51 is: https://www.oasis-open.org/apps/org/workgroup/camp/download.php/50490/camp-spec-v1.1-issue-51-v1.doc
Gil Pilz (Oracle): Jacques: we need to fix the pseudo-schema for 4.3.6
Gil Pilz (Oracle): Ashok: there is no JSON schema standard, so we are sort of creating an informal JSON schema
Gil Pilz (Oracle): ... there are corner cases etc. that we should specify formally
Gil Pilz (Oracle): ... there is a JSON schema effort called "JSON-LD"
Jacques (Fujitsu): the scoping of  ? * + seems implicitly to be the entire preceding line. This "default" should be explicitly stated. Also, the scoping of "|" in 4.3.6 is not clear . Maybe thats where () should be used to scope the alternative.
Adrian Otto (Rackspace): Gil indicates that the pseudo-code we offer acts as a synopsis for the text, and the text is the actual spec.
Gil Pilz (Oracle): Adrian: Ashok, could you review JSON-LD?
Anish Karmarkar: sorry, we got disconnected
Gil Pilz (Oracle): Ashok: yes, might take a bit
Anish Karmarkar: tom pressed the wrong button, will be on in a few secs
Gil Pilz (Oracle): Tom: I indicated (above) the wording for indicating that the default cardinality is 1
Gil Pilz (Oracle): Anish: I'm fine with the text that Tom is suggesting
Gil Pilz (Oracle): ... I like pseudo-schema
Gil Pilz (Oracle): ... regardless of whether we add information based on JSON-LD, I would like to maintain the pseudo-schema
Gil Pilz (Oracle): ... symbols are familiar, and text takes precedence
Gil Pilz (Oracle): Adrian: I support the use of pseudo schema
Gil Pilz (Oracle): ... I'm not sure how brief JSON-LD
Gil Pilz (Oracle): Gil: will turn this around so we can discuss next week
Gil Pilz (Oracle): Tom: whoever is investigating JSON-LD, it looks like it is tied to ontologies and RDF tuples

TOPIC: Labeling of Normative Statements is incomplete (aka overhaul V2) http://tools.oasis-open.org/issues/browse/CAMP-83

Gil Pilz (Oracle): Jacques: CAMP-83 is an umberella issue for a number of different concerns
Gil Pilz (Oracle): ... I did some work on it last week
Gil Pilz (Oracle): ... would like to review the current "challenges"
Gil Pilz (Oracle): ... we may want to leave the issues around PDP-20 open
Gil Pilz (Oracle): ... the idea is that future specs may profile or use CAMP in some way that places meaning on the order of artifacts in the array, but CAMP itself doesn't do so
Adrian Otto (Rackspace): How are we in terms of progress toward goal of completing TA's by close of public review? Are we ahead or behind schedule?
Gil Pilz (Oracle): Jacques: we are on schedule
Gil Pilz (Oracle): ... would be OK with approving CAMP-83 today
Gil Pilz (Oracle): Gil: would like a chance to tweak some of the wording
Gil Pilz (Oracle): Jacques: would like to review some of the challenges
Gil Pilz (Oracle): .. RE-16
Gil Pilz (Oracle): Because of the unique function of this resource, future versions of the CAMP specification SHALL NOT make non-backwards compatible changes to this resource.
Gil Pilz (Oracle): Gil: we're trying to promise developers that they will have this bootstrap mechanism to figure out what versions a provider supports
Gil Pilz (Oracle): Mike: is this a way to query what versions are available?
Gil Pilz (Oracle): Adrian: this is advice to future members of the TC about the fact that you can't change this in a non-backward compatible way
Gil Pilz (Oracle): ... we should do this, but a normative statement is not the way to do this
Gil Pilz (Oracle): ... might just use a note
Gil Pilz (Oracle): Mike: sounds good to me
Gil Pilz (Oracle): Jacques: this specification cannot constrain what the next version will do
Gil Pilz (Oracle): ... but there is an implicit commitment not to break this resource in particular
Gil Pilz (Oracle): ... none of this is testable
Gil Pilz (Oracle): Jacques: RE-01, RE-03, RE-04
Gil Pilz (Oracle): ... these requirements seem redundant with the notation that we use to describe attributes in general
Gil Pilz (Oracle): ... they seem to have been put in place before we established a consistent usage of that notation
Gil Pilz (Oracle): Gil: we haven't established that notion at the point where RE-01 occurs, looks like an editorial problem
Gil Pilz (Oracle): Jacques: (agrees)
Gil Pilz (Oracle): Jacques: RE-55, RE-56
Gil Pilz (Oracle): (misc. discussion)
Gil Pilz (Oracle): If this attribute is present in an instance of this resource, Providers SHALL support the POST method on that instance in addition to the methods defined in Section 5.5, ?HTTP Method Support <#_Platform_Resource>?. [RE-38]
Adrian Otto (Rackspace): Note that RE-56 needs the "instance" terminology removed from it.
Gil Pilz (Oracle): Jacques: yes, we need to clean up the use of "instance"
Gil Pilz (Oracle): ... about all we can say CAMP-83

Gil Pilz (Oracle): Gil: I'll make my changes and we can interate over that

TOPIC: State change mechanism is unclear http://tools.oasis-open.org/issues/browse/CAMP-31

Gil Pilz (Oracle): Gil: do we want to define a set of canonical application states in CAMP (obviously extensible)?

Gil Pilz (Oracle): ... if so, how should we represent that info in our resource model?

Gil Pilz (Oracle): Jacques: there is a 3rd option - don't define canoncial states, but define the way in which this undefined state is represented

Gil Pilz (Oracle): Derek: we've learned, through years and years of drudgery, that you do end up with a common vocabulary of states, but the transitions are quirky

Gil Pilz (Oracle): ... what seems to work is defining different lifecycles that work

Gil Pilz (Oracle): ... we have about 5 that we actually use

Gil Pilz (Oracle): ... things get pretty confusing if you don't have something like this

Gil Pilz (Oracle): Ashok: how are you going to specify a lifecycle without spelling out the states?

Gil Pilz (Oracle): ... don't you have to specify where you start, what the next states might be, how you traverse between them?

Gil Pilz (Oracle): Derek: the possible vocabulary of states that you can put in the diagram is more or less fixed

Gil Pilz (Oracle): ... but the transitions vary between different scenarios

Gil Pilz (Oracle): ... for example, web apps versus ???

Gil Pilz (Oracle): ... state names are the same, but edges vary between different kinds of apps

Gil Pilz (Oracle): Derek: whether you are modeling the app or modeling the components, you end up drawing the diagram and sharing it with people

Gil Pilz (Oracle): ... if we want people to consume the apps we model, we have to transmit this to them somehow

Gil Pilz (Oracle): ... basically by passing URLs to documentation

Gil Pilz (Oracle): TC is roughly in favor of  defining some minimal/core set of application states

Gil Pilz (Oracle): Ashok: would like to see what Derek sends us

TOPIC: Straggler Roll

Gil Pilz (Oracle): Derek Palma

MEETING ADJOURNED