Re: [opencsa-liaison] Reference dependencies in various SCA specs

[Anish Karmarkar <Anish.Karmarkar@oracle.com>]

Paul,

Thanks for correcting that. Should have been more precise in my email.
But in the context of the current discussion, the SCA TCs are looking to 
go to CS soon. The problem they face is that to go to CS, the current 
rules will require infinite number of CSDs and PRs because of the 
cross-dependencies, unless we find a way to break the cycle. I have 
proposed three different ways to move this forward. Perhaps there are more.

-Anish
--

On 8/8/2011 7:37 PM, Paul Knight wrote:
> Hi all,
>
> A point which I think may be misquoted and/or misunderstood:
>
> "the requirement of new PRs for any change"
>
> mentioned by both Jeff and Anish is not  a "requirement".
>
> A PR is needed ONLY if the immediately following step is a vote for
> making the Work Product a Committee Specification (or Note).  TCs should
> not believe that they have to do a new PR every time they produce a CSD.
>
> In fact, avoiding doing the PR until all the cross-referenced Work
> Products are at stable CSD states may provide a known and fairly stable
> target URI (i.e, the upcoming PR version), which may allow some
> efficiencies for everyone.
>
> However, in any case, Chet is setting up a process to get concerns and
> approaches clarified.
>
> Best regards,
> Paul
>
> On Mon, Aug 8, 2011 at 10:01 PM, Jeff Mischkinsky
> <jeff.mischkinsky@oracle.com <mailto:jeff.mischkinsky@oracle.com>> wrote:
>
>     we can also allocate some time in Process Committee if that would be
>     desireable/easier.
>
>     FWIW I think the problem is that the SCA TCs are not really
>     completely independent of each other. (I think of them as "almost
>     separable".) When we were first developing SCA we discussed whether
>     we should have one big TC or break things up to make development
>     more manageable realizing that the trade-off was that we would have
>     mutual dependencies.
>
>     That said i'm not sure the real issue is independent TCs. The mutual
>     dependencies would exist even if there was one TC. Having multiple
>     TCs makes it more complicated because its harder to coordinate
>     simultaneous votes on mutually dependent specs across TCs. I'm
>     trying to remember why we don't allow DCR's for CSD's. I *think* it
>     might be because we thought that since these were "only drafts", it
>     was ok if there were some inconsistencies/bugs and not worth the
>     effort to allow DCR's.
>
>     The suggestion to allow TC's to decide if a change is not
>     "substantive enough" is that we were never able to agree on a
>     definition of what "enough" meant and TCs will almost always declare
>     a change to be non-substantive, regardless.
>
>     Actually we already have that issue on the Process Committee's
>     agenda as part of figuring out if the requirement of new PRs for any
>     change is creating too many PRs.
>
>     cheers,
>       jeff
>
>
>     On Aug 08, 2011, at 5:14 PM, Chet Ensign wrote:
>
>         Mike, Sanjay - Actually, before launching on a broad chat, I
>         think it would make more sense for us to talk and frame the
>         problem to be solved so that conversation can be productive. Can
>         I set up a call with both of you, maybe for Wednesday or
>         Thursday? That will give me some time to draft a first pass at a
>         problem statement that we can use to set up productive
>         discussion with others.
>
>         /chet
>
>         On Mon, Aug 8, 2011 at 6:40 PM, Patil, Sanjay
>         <sanjay.patil@sap.com <mailto:sanjay.patil@sap.com>> wrote:
>         I think it makes a lot of sense to have a conference call to
>         jointly define the problem and brainstorm the potentially
>         solutions. However, scheduling a call with the many interested
>         parties from the various time zones might become tricky. One
>         option may be to check with the SCA Assembly TC chairs if they
>         would be ok to allocate some of their weekly meeting time (Tue 8
>         AM PST) for this topic (assuming that Chet and other concerned
>         OASIS staff / TAB members can make it). I believe that most of
>         the SCA TC chairs are also members of the SCA Assembly TC and
>         moreover the SCA TCs typically look forward to the SCA Assembly
>         TC to resolve such cross-cutting issues and set the precedent.
>         Mike/Martin, any thoughts in this regard?
>
>
>
>         Sanjay
>
>
>
>         From: Chet Ensign [mailto:chet.ensign@oasis-__open.org
>         <mailto:chet.ensign@oasis-open.org>]
>         Sent: Monday, August 08, 2011 3:20 PM
>         To: Patil, Sanjay
>         Cc: Mike Edwards; Anish Karmarkar; OASIS Liaison; OASIS TAB;
>         Paul Knight; Robin Cover
>         Subject: Re: [opencsa-liaison] Reference dependencies in various
>         SCA specs
>
>
>
>         Sanjay, Mike et al - We started a discussion on this subject at
>         the TAB F2F and it is on our agenda. If you all would like, I
>         would like to schedule a conference call so that we can discuss
>         together and ...
>
>
>
>         - Jointly define the problem(s) - typically the most difficult
>         part of any of these conversations <grin>
>
>         - Discuss what limitations of the current process / rules
>         contribute to the problem(s)
>
>         - Brainstorm possible modifications that could resolve the
>         problem(s)
>
>
>
>         Is that something you all are interested in doing?
>
>         Best,
>
>
>
>         /chet
>
>         On Mon, Aug 8, 2011 at 5:05 PM, Patil, Sanjay
>         <sanjay.patil@sap.com <mailto:sanjay.patil@sap.com>> wrote:
>
>
>
>         Do other TC chairs have any further viewpoints on this topic?
>
>
>
>         Since the resolution of this issue may involve changes to
>         procedures involving the TC Admin (and potentially the TC
>         process!), we need to decide how to take this issue forward. At
>         the minimum, I think we should inform the TC Admin about our
>         opinions and start the discussion. I am taking the liberty to
>         copy Chet Ensign on this email thread to set the ball rolling.
>
>
>
>         Best wishes,
>
>         Sanjay
>
>
>
>         From: Mike Edwards [mailto:mike_edwards@uk.ibm.__com
>         <mailto:mike_edwards@uk.ibm.com>]
>         Sent: Wednesday, August 03, 2011 1:33 AM
>         To: Patil, Sanjay
>         Cc: Anish Karmarkar; OASIS Liaison
>         Subject: RE: [opencsa-liaison] Reference dependencies in various
>         SCA specs
>
>
>
>
>         Sanjay,
>
>         I agree with your views.
>
>         In the latest round of changes, for example, I think that
>         Assembly is not affected by the changes to the
>         other specifications (Policy, WS-Binding, Java...), even though
>         those changes are normative.
>
>         As a result, in my opinion, updating the Assembly spec to
>         reference the latest CSDs of the other specs
>         is a purely editorial process.
>
>         Each TC may need to take a vote that agrees this for any
>         particular spec, but once that vote has
>         taken place, the update should be purely mechanical.  No new
>         public reviews are required.
>
>
>         Yours, Mike
>
>         Dr Mike Edwards
>
>           Mail Point 137, Hursley Park
>
>         <image001.gif>
>
>
>         STSM
>
>           Winchester, Hants SO21 2JN
>
>         SCA, Cloud Computing & Services Standards
>
>           United Kingdom
>
>         Co-Chair OASIS SCA Assembly TC
>
>
>
>         Chair UK ISO SC38 mirror committee (Cloud & SOA)
>
>
>
>         IBM Software Group
>
>
>
>         Phone:
>
>         +44-1962 818014 <tel:%2B44-1962%20818014>
>
>
>
>         Mobile:
>
>         +44-7802-467431 <tel:%2B44-7802-467431> (274097)
>
>
>
>         e-mail:
>
>         mike_edwards@uk.ibm.com <mailto:mike_edwards@uk.ibm.com>
>
>
>
>
>
>
>
>
>
>
>         From:
>
>         "Patil, Sanjay" <sanjay.patil@sap.com <mailto:sanjay.patil@sap.com>>
>
>         To:
>
>         Anish Karmarkar <Anish.Karmarkar@oracle.com
>         <mailto:Anish.Karmarkar@oracle.com>>, OASIS Liaison
>         <opencsa-liaison@lists.oasis-__open.org
>         <mailto:opencsa-liaison@lists.oasis-open.org>>
>
>         Date:
>
>         02/08/2011 17:56
>
>         Subject:
>
>         RE: [opencsa-liaison] Reference dependencies in various SCA specs
>
>
>
>
>
>
>
>         I think the option 3 provides the required solution while
>         retaining the independence for the TCs to decide and follow
>         their own schedule. During CSD publication and PR, specs can
>         make references to the existing source material as of that time.
>         Later, it should be allowed to update the references without
>         having to go for another round of PR. I think it is reasonable
>         to trust the TCs in making a determination about whether a
>         change in a reference is significant enough to require another
>         round of PR or not.
>
>         Sanjay
>
>
>         -----Original Message-----
>         From: Anish Karmarkar [mailto:Anish.Karmarkar@__oracle.com
>         <mailto:Anish.Karmarkar@oracle.com>]
>         Sent: Tuesday, August 02, 2011 9:22 AM
>         To: OASIS Liaison
>         Subject: [opencsa-liaison] Reference dependencies in various SCA
>         specs
>
>         As we all know, specs in all the SCA TCs have dependencies on each
>         other. For example, Assembly has a normative reference to Policy and
>         WS-Binding. To conform to Assembly one has to conform to Policy and
>         WS-Binding. Policy in turn depends on Assembly. WS-Binding
>         depends on
>         Assembly/Policy. The dependency graph is cyclic.
>
>         A similar problem exists in SCA-J TC but without any cycles [1].
>
>         OASIS TC process solves this problem of cyclic references by using
>         designated cross references [2]. Unfortunately, CSDs cannot use this
>         mechanism as the process explicitly excludes it. So, two specs A
>         and B
>         that depend on each other, will be in an infinite loop wrt updating
>         their normative references. Further compounding the problem (but
>         is an
>         independent issue) is the fact that the new process requires at
>         least
>         15-days PR for any change (except the changes on the front
>         page). So any
>         time spec A or B is updated to update its normative reference it
>         has to
>         be PRed.
>
>         I see three possible solutions to this:
>         1) Fix the process to allow DCRs for CSDs (this is the only long
>         term
>         solution, I think).
>         2) Use the solution in [1] where spec A instead of using WDXX
>         reference
>         of B, uses the CSDYY reference of B. Where CSDYY is the what
>         will happen
>         to WDXX when the TC Admin processes the request for WDXX to be
>         published
>         as CSD. This means that the CSDs of A and B should be
>         coordinated to go
>         in the TC Admin queue around the same time.
>         3) Get TC admin to update normative references when there is no
>         material
>         change. For example, in (2) above, WDXX and CSDYY are going to be
>         identical specs except for the front page. If spec A references
>         WDXX,
>         the TC admin should be able to update that reference to CSDYY. A
>         similar
>         update would have to be done to spec B.
>
>         -Anish
>         --
>
>         [1]
>         http://lists.oasis-open.org/__archives/sca-j/201107/__msg00028.html
>         <http://lists.oasis-open.org/archives/sca-j/201107/msg00028.html>
>         [2]
>         http://www.oasis-open.org/__policies-guidelines/tc-__process#crossRefs
>         <http://www.oasis-open.org/policies-guidelines/tc-process#crossRefs>
>
>         ------------------------------__------------------------------__---------
>         To unsubscribe from this mail list, you must leave the OASIS TC that
>         generates this mail.  Follow this link to all your TCs in OASIS at:
>         https://www.oasis-open.org/__apps/org/workgroup/portal/my___workgroups.php
>         <https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php>
>
>
>
>
>
>
>
>         Unless stated otherwise above:
>         IBM United Kingdom Limited - Registered in England and Wales
>         with number 741598.
>         Registered office: PO Box 41, North Harbour, Portsmouth,
>         Hampshire PO6 3AU
>
>
>
>
>
>
>
>
>         --
>
>         /chet
>         ----------------
>         Chet Ensign
>         Director of Standards Development and TC Administration
>         OASIS: Advancing open standards for the information society
>         http://www.oasis-open.org
>
>         Primary: +1 973-378-3472 <tel:%2B1%20973-378-3472>
>         Mobile: +1 201-341-1393 <tel:%2B1%20201-341-1393>
>
>         Follow OASIS on:
>         LinkedIn: http://linkd.in/OASISopen
>         Twitter: http://twitter.com/OASISopen
>         Facebook: http://facebook.com/oasis.open
>
>
>
>
>         --
>
>         /chet
>         ----------------
>         Chet Ensign
>         Director of Standards Development and TC Administration
>         OASIS: Advancing open standards for the information society
>         http://www.oasis-open.org
>
>         Primary: +1 973-378-3472 <tel:%2B1%20973-378-3472>
>         Mobile: +1 201-341-1393 <tel:%2B1%20201-341-1393>
>
>         Follow OASIS on:
>         LinkedIn: http://linkd.in/OASISopen
>         Twitter: http://twitter.com/OASISopen
>         Facebook: http://facebook.com/oasis.open
>
>
>     --
>     Jeff Mischkinsky jeff.mischkinsky@oracle.com
>     <mailto:jeff.mischkinsky@oracle.com>
>     Sr. Director, Oracle Fusion Middleware +1(650)506-1975
>     <tel:%2B1%28650%29506-1975>
>             and Web Services Standards                              500
>     Oracle Parkway, M/S 2OP9
>     Oracle
>       Redwood Shores, CA 94065
>
>
>
>
>
>
>
>
>
>
>
>
> --
> Paul Knight <mailto:paul.knight@oasis-open.org>  - Tel: +1 781-861-1013
> OASIS <http://www.oasis-open.org/> - Advancing open standards for the
> information society
> Document Process Analyst
> <http://www.oasis-open.org/people/staff/paul-knight>
>
>