See also: IRC log
<scribe> Scribe: Len Charest
<scribe> ScribeNick: lencharest
<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2009Jan/att-0025/20090115-sml-minutes.html
RESOLUTION: Minutes approved
<scribe> No progress, but the required updates to speadsheet have been made by others
John: Will close action 215
MSM: Have draft wording for old namespace docs
<MSM> http://www.w3.org/2008/09/sml
<MSM> http://www.w3.org/2008/09/sml-if
MSM: Above URIs show current wording
Len: Change all occurences of 'namespaces' to 'namespace'
Otherwise, Group approves draft language
MSM: Webmaster put them under ns/ directory, along with namespace docs
... But I see no other schema docs in the ns/ directory
... Propose that we adopt policy of publishing "dated" schema URIs akin to the dated URIs for TRs
RESOLUTION: Dated schema URIs will be implemented
John: Propose 12:30pm-1:30pm Eastern on Mondays starting with first meeting in Feb
RESOLUTION: WG teleconference will take place from 12:30pm-1:30pm Eastern on Mondays starting with first meeting in Feb
RESOLUTION: Text will be changed per description in bug text
John: Bug updated as editorial
http://www.w3.org/Bugs/Public/show_bug.cgi?id=6245
John: Looks like we agreed on this in December, now that 5680 is editorial 6245 can be done as well.
John: Absent other comments, let us address the questions posed already in email:
<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2009Jan/0041.html
Current rev of spreadsheet in CVS is 1.6
Ginny: derefNV is about testing smlfn:deref outside a validator; will we have an implementation
John: Depends on how we define the test and/or the feature
... One option is to say derefNV is not relevant for demonstration of implementation experience
... Option two -- change the definition of derefNV so that we don't need to test outside the context of a validator
Keep in mind that the test suite purpose is to show evidence of implementability, not compliance,
as we have discussed at the F2F.
Ginny: Focus of tests is on SML validators
John: Conclusion seems to be that derefNV is not relevant for SML interop
Kumar: SML spec section 4.2.7 has note adressing smlfn:deref and interop
MSM: There is value in separation of defn of smlfn:deref from SML validation; smlfn:deref is useful outside of validation
... reluctant to lose the derefNV feature
Ginny: Would COSMOS serve as an implementation of derefNV because of how smlfn:deref is implemented?
Sandy: Apologies: just realized I have another meeting at 3 that I cannot miss, so I need to leave very soon. I see a PR decision on the agenda for later today - so that need not be deferred, I am fine with requesting a transition to PR if the rest of the wg agrees. Bye for today.
John: COSMOS does implement Schematron rule checking (rules may contain deref() calls)
by invoking Xalan (an open source XSLT processor) on the Schematron stylesheet from Rick Jelliffe after
setting up the XPath function library context entry for smlfn:deref.
... We have consensus to update the spreadsheet as described by http://lists.w3.org/Archives/Public/public-sml/2009Jan/0022.html
Kumar will make change #1 -- adding a second expected results column to differentiate between implementations supporting the
feature and one for those not supporting it
John: Second email containing some questions we need to discuss wrt impl report
<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2009Jan/0047.html
John: Text re row 19 actually applies to row 16
Kumar: MS impl matches expected result for row 16
John: Re row 29 -- Multiple sml:uris means not a valid SML ref; should we remove the test case?
Kumar: suggest updating expected result instead
Ginny: If test is discarded, then create a test that specifically addresses non-root elements
<johnarwe_> <sml:uri>Courses#smlxpath1(/u:Courses/u:Course[2])</sml:uri>
John: There are many other tests that target non-root; shall we remove row 29?
MSM: Let's just mark it invalid
John: Seems confusing - in the past, we've removed invalid tests like multiple sml:uri elements inside a single SML ref... e.g. old row 31, now blanked out. Removing them from our list also doesn't stop anyone, e.g. cosmos, from running them. Cosmos actually will continue to run all their tests regardless of what SML does, via automated JUnits.
MSM: Curious that I was ok with removing them before. I can understand a desire for consistency, so OK either way (marking invalid or removing).
John:
For consistency, let's remove the test in row 29 (but preserve the row so that row numbers remain the same)
... Re row 76 --- need MS results
Kumar: MS results are 76 is false, 77 is true
John: Re 96-99 --- we should add remarks to suggest cosmos alter the tests so a non-supporting impl would find the model invalid
... How shall we define test pass? Percentage of tests where actual result equals expected result?
MSM: W3C has no rule on this, simply need to be persuasive enough to convince the Director. Current results (1-2 wrong results out of ~150 tests) seem likely to be persuasive.
MSM: We need a "clean" copy of the spec that is ready to publish at any time
... We need an estimate from the editors on when we will have a ready-to-publish draft
John: Shall we request transition to PR?
... Sandy has already said "yes"
... No objections heard, so we have consensus
RESOLUTION: WG will request transition to PR status for SML and SML-IF specs
Ginny: Outstanding editorial changes will be made within a week
<johnarwe_> http://lists.w3.org/Archives/Public/public-sml/2009Jan/0022.html
<johnarwe_> under "changes still needed"
<johnarwe_> all have 2x - (--) in front
Last Scribe Date Member Name Regrets pending 2008-05-22 Lynn, James Until further notice 2008-10-29 Pandit, Kumar 2008-12-11 Wilson, Kirk 2009-01-08 Smith, Virginia 2009-01-15 Gao, Sandy 2008-01-22 Charest, Len Exempt Arwe, John Exempt MSM