W3C SML Teleconference -- 11 Sep 2008

<scribe> scribe: Kirk Wilson

<scribe> scribenick: Kirk

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Sep/att-0007/20080904-sml-minutes.html

Approval of minutes of 9/4

No objections to approving minutes as true record of meeting of 9/4

From CG

John: People should register for Oct meeting.
... vote concerning February meeting now closed.

Action Items

Action 137: Kumar to talk to Pratul

Kumar: Agreed with Pratul do not have strong feelings about making the request to the schema group regarding inheritance of identity constraints. But no objection to it.
... Microsoft does not want to reconsider the original decision to write a proposal to the Schema group.

John: Action assigned by the wg was to write a proposal. It does not appear that this action was executed.
... Working group must reassign action or reconsider its decision.
... Is there anyone else who would like to write the proposal? (Do we impede Schema 1.1 by submitting past deadline of LC comments?)

MSM: Important to get on Schema's agenda. Would like someone not also a member of the schema group to write it, to avoid any possible perception of a personal agenda being driven. But MSM is willing to write it so that it doesn't fall off the agenda.

John: Would full endorsement by the WG help to alleviate concern?

MSM: Yes, will open bug with indication an that the SML WG will or will not endorse it later, in order to get it on the agenda before the close of Schema's LC review period.

John thanks MSM for filling in on this task.

John is reassign action 137 to MSM to open bugzilla issue.

MSM: Hope to complete action 190 this afternoon. Develop sample for schema validity assessment.

Endorsement of Schema bugs openned by John

<johnarwe_> http://www.w3.org/Bugs/Public/show_bug.cgi?id=6009 through http://www.w3.org/Bugs/Public/show_bug.cgi?id=6015

Unclear Passages: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6009

MSM: Proposes that we endorse 6009 as coming from the entire SML WG

RESOLUTION: The WG endorses 6009.

Priority Feedback responses: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6010

John: This responds to specific to requests from the Schema 1.1 editors for feedback.

MSM: Endorse 3.3.4.6 and the last one on redefine.
... Comfortable with endorsing all five, but middle 3 do not deal with SML issues.

Issue 1. 3.3.4.6 Fall back to Lax

RESOLUTION: No objections to endorsement.

Kumar: regarding Issue 1, What does "fallback" mean?

MSM: For wildcard for which you have no declaration, should validator say, "I won't validiate" or "I validate subtree as far as possible". We prefer to fallback to lax, i.e., validate subtree.

Issue 2: 3.3.4.5 (not part of SML).

RESOLUTION: No objections to endorsement

Issue 3: 3.10.1 ##Define keyword (not part of SML)

RESOLUTION: WG endorses comment without objection.

Issue 4: 3.10.1 ##Sibling keyword (not part of SML).

RESOLUTION: WG endorses comment without objection.

Issue 5: 4.2.3 Redefine

RESOLUTION: WG endorses comment without objection.

John: Adds note in "SML endorses..." comment on 6010 that SML no longer mentions "redefine" but member submission excluded it.

Base URI comments: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6011

Kirk: Propose that we endorse 6011

RESOLUTION: WG endorses 6011 without objection.

Inconsistencies in text: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6012

Kirk: Propose that we endorse 6012

RESOLUTION: WG endorses 6012 without objection.

New examples needed: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6013

No proposal. WG did not endorse this issue.

Normative text problems: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6014

MSM: Marked as editorial = passages will be recast to normative or rewritten without normative implications.
... These will be acted upon on matter what the SML WG does.

RESOLUTION: No objections to "punting" on 6014 = not choosing to endorse it (and not choosing to not endorse it, i.e. SML chooses not to decide rather than spend any time, since any hypothetical SML endorsement would not change the bug's handling).

Valid (and its poor relations) vs. assessment (and its...): http://www.w3.org/Bugs/Public/show_bug.cgi?id=6015

John: There is ample opportunity to confuse readers, also possible contradictions.

Kirk: Proposes endorsement of 6015

RESOLUTION: WG endorses 6015 without objection.

Test Case Discussion.

<johnarwe_> Emails on interop test plan, in date order

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Apr/0062.html Kumar 1st draft

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Apr/0067.html Ginny comments

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Apr/0069.html Kumar response + 2nd draft (1st+line #s)

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Apr/0073.html Ginny (brief) comments

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008Apr/0087.html Kirk "what is interop[able]?"

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008May/0017.html John comments on 2nd draft attached

<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2008May/0023.html John "need to know features"

Kumar: We should discuss whether we agree with the test plan.

Kumar: 'the' meaning http://lists.w3.org/Archives/Public/public-sml/2008Apr/0069.html

MSM: Comment on p.1 lines 6-8 Overview: The criteria are minimal criteria, they are not exit criteria until chairs and Director agree to these. Groups have criteria higher than these.
... If we walk in with one implementation of each Required feature, there will not be a successful Director's call.
... Concrete revision to this sentence. (MSM will draft the revision.)

<MSM> In order to progress from Candidate Recommendation status to Proposed

<MSM> Recommendation, any W3C specification must satisfy the exit criteria

<MSM> for Candidate Recommendation agreed upon by the Director and the

<MSM> chairs of the responsible Working Group. At a minimum, the W3C

<MSM> process document requires that the specification have at least one

<MSM> implementation of each feature, and preferably two inter-operable

<MSM> implementations; additional exit criteria may be (and often are)

<MSM> agreed by the Director and chairs.

<johnarwe_> last line: agreed to by ?

Kumar: This should say what we intended to do.

MSM: Exit criteria are agreed upon by Director and chairs. We have to be in position to provide the appropriate information.

<MSM> In order to progress from Candidate Recommendation status to Proposed

<MSM> Recommendation, SML and SML-IF must satisfy the exit criteria for

<MSM> Candidate Recommendation agreed upon by the Director and the chairs of

<MSM> the SML Working Group. At a minimum, the W3C process document

<MSM> requires that each specification have at least one implementation of

<MSM> each feature, and preferably two inter-operable implementations;

<MSM> additional exit criteria may be (and often are) agreed by the Director

<MSM> and chairs. This document defines the approach adopted by the SML

<MSM> working group to meet that goal.

MSM: We cannot say (now) what is likely to be the exit criteria (decided later in time). We will need to discuss what criteria the chair will take to the Director. Test Plan must be able to provide the information regarding what features have been tested.
... It is premature to talk about exit criteria in this document. We need to have a test plan to identify what features are tested, not the number of implementations.

<MSM> As a consequence, the basic requirement for our test plan is that the

<MSM> SML WG must construct a test suite suitable for documenting (a) which

<MSM> features of the spec have been implemented and (b) for each feature

<MSM> implemented more than once, whether the implementations are consistent

<MSM> and interoperable. This document defines the approach adopted by the

<MSM> SML working group to meet that goal.

Kumar: Agrees that this "makes sense".

<johnarwe_> Pratul, we are discussing http://lists.w3.org/Archives/Public/public-sml/2008Apr/att-0069/sml-interop-test-plan.pdf

No objections to revisions to section 1 as proposed by MSM.

Section 2: Test packaging.

MSM: We have to be able to test two SML-IF documents with documents in-line or not. We may find it more convenient to have a catalog of tests. A third possibility: test description in a catalog where each test is test described with metadata.
... We cannot use either method is isolation because we have to test in-line and not in-line document.
... This approach contradicts p. 2 lines 17-18.
... We need to have tests for the locator element.
... We need to specify behavior when an optional feature is not supported.

Section 3: Test Storage and Organization

MSM: If we plan to have a test catalog, we should say in the directory structure where catalogs will go. Also, where test result files will be placed.

John: Are progress reports typically stored in CVS?

MSM: Typically, Yes.
... Some groups make provision for anonymous submission of test results.

John: Does group realize we may have pointers to other test plans, e.g., COSMOS?

<johnarwe_> I was concerned with ptrs to test cases, not plans.

Pratul raises issue of persistence of these cases/durability of these references, i.e. that the pointed-to content may be changed after the wg agrees on it. Need to be able to point to a 'frozen' point in time copy.

MSM: Nothing requires that all test suites must reside on W3C servers. But we should have catalogs in one place.

Section 4: Test Execution

MSM: If we decided to have a test result format, should this go into this section?

Section 5: Analyzing Test Results

Pratul: First line: WG is not required to have anything.

MSM: "As mentioned earlier, the test suite will have at least one test for every feature."
... Revision to second paragraph, lines 9 -12.
... Second question: If processors implement an optional feature differently, then question is why. There are two results: (1) don't implement the optional feature, (2) 2 processors should implement the optional feature in the same way. We should expect them to implement the optional feature in the same way .
... Say the same about two implementations that DON'T implement a feature. If you don't support feature X, then you get result Y. Two impls with identical feature sets should exhibit the same behavior.

Pratul: We should not spend a great of time perfecting this document because we expect only two implementations.

Meeting adjourned: 4:02 ET.

W3C SML Teleconference

11 Sep 2008

Attendees

Contents

Approval of minutes of 9/4

From CG

Action Items

Endorsement of Schema bugs openned by John

Test Case Discussion.

Summary of Action Items

Updated ScribeList