See also: IRC log
<scribe> Scribe: Charlie Wiecha
<scribe> scribenick: wiecha
<ebruchez> FYI will be a few mins late
John: is it easy for you (Nick) to create updates to your test report?
Nick: yes, have a couple of stylesheets to do the merge of results and then generate HTML
John: for Ubiquity, have asked
folks to create issues for features we're not passing but are
orange/red in your report...meaning we don't have enough impls
to proceed...not that many exist
... issue list should be coming soon
... mailto might be hard
Nick: I can look at that since
we've done that in Chiba
... but it's pretty busy here now
John: we should be able to do an implementation report from Ubiquity soon
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0025.html
John: talked with Nick about what
we have in CVS now...most folks working on it have access
now
... there's lots of licensing stuff to be worked out if we want
to go external
... would need CLAs etc for participants
... probably best to just keep as it is for now
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0035.html
this is in a subsection of the spec, whose removal would cause renumbering of the spec
John: have massaged text to avoid this so that impl reports etc are not impacted
<scribe> ACTION: Nick to update spec to correct section numbering given link error removal [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action01]
<trackbot> Created ACTION-527 - Update spec to correct section numbering given link error removal [on Nick Van Den Bleeken - due 2009-04-01].
John: timing not critical so appreciate you can get to it when possible
Nick: normally, yes
... it's an ant build
John: Erik -- have you been able to update output w/label issue etc?
Erik: will look at this this week
John: Nick, can you point to the ant script?
Nick: yes, it's in CVS
John: your style sheets etc that creates analysis across impls?
Nick: right, that's not checked in yet
John: if you could put that in the 2008 directory under the impl reports that would be nice
Nick: those files can also be externally downloaded...is that ok?
John: yes, tools for analysis are available too
which is ok
<nick> /Forms WG/Test/XForms1.1/Edition1/zip/build.xml
John: might help other people run the builds, and also perhaps to include our results in other test suites
Nick: ok, will check them in
<scribe> ACTION: Nick to check in ant scripts for building test results into CVS [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action02]
<trackbot> Created ACTION-528 - Check in ant scripts for building test results into CVS [on Nick Van Den Bleeken - due 2009-04-01].
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0031.html
John: made things strange to say
impls could do either approach, so I would up saying it's lower
case
... tests were invoking translate to force lowercase, which we
don't technically need now...probably doesn't matter to keep
this
... but would be nice to remove this translation
<scribe> ACTION: John_Boyer to update tests to change digest and hmac tests to remove translation to lowercase [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action03]
<trackbot> Sorry, couldn't find user - John_Boyer
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0036.html
John: we said actions on root
(insert/delete) did not have defined behavior here
... when checking the spec, setvalue already had defined
semantics there...so didn't change setvalue
... what it says however is different from insert/delete
... setvalue says namespace node targeted action is
undefined
... vs ignored for insert/delete
... leave implementors a bit of wiggle room for setvalue
... which seems like a better idea so left unchanged
... setvalue on root node produces binding exception
... if you disagree, please post to the list
http://lists.w3.org/Archives/Public/public-forms/2009Feb/0091.html
John: problem is
xforms-link-error is used to test delete method of
submission
... question to the group is how to modify the test to still
test delete?
... Nick, do you remember is this local file or remote
resource?
Nick: rel path to the test file beside the form...then load trigger which should produce link error to missing file
John: now we say load indicates failure in implemention-specific way
Nick: right, so not clear how to update the test
John: rather than using link-error, could the test just try to activate the load file trigger, should fail to load?
Nick: yes
John: then just delete the action that listens for the the link error
Paul: or use second submission that tries to load and then look for submit error there
John: better since that can be automated
Nick: it's now a text file but we can use an xml file
John: on replace="all" we don't
get a submit error
... btw, not sure we have any impls passing the test getting
done or error events on replace="all"
... will have to deal with this across all impl reports
... we could replace text in the same page to see submit
error
... to fix this w/o needing replace="all"
Nick: but then we need another 1.1 feature to test this
John: but delete is a 1.1 feature
Nick: I haven't implemented replace text
John: then we could use xml and replace instance instead
Nick: ok
John: but whatever file we use is
going to be deleted anyway :}
... is there a specific error we should look for?
... looks like a resource error
... Nick, could you modify this test?
Nick: sure, should we fill in response code? or is this too much function?
John: probably ok to just check submit-error
<John_Boyer> <message ev:event="xforms-submit-error" if="event('error-type')='resource-error'">You pass</message>
<scribe> ACTION: Nick to update test 11.9.4.b to test for xforms-submit-error [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action04]
<trackbot> Created ACTION-529 - Update test 11.9.4.b to test for xforms-submit-error [on Nick Van Den Bleeken - due 2009-04-01].
http://lists.w3.org/Archives/Public/public-forms/2009Feb/0084.html
John: don't know if we have a
test for this yet
... yes, there is one
... we state it's possible to use a submission to validate
data
... without needing to actually transmit to the server if
valid
... use case is just validation
... not elegant but works for 1.1
... so we should have a test for this
... if submission has not resource or action attribute or
element then submission will fail but the rest of the pipeline
gets executed
... so it should be possible to use submission and just look
for failing with validation error vs resource error
... and just ignore the resource error
... any objections to add this to the test suite?
+1
<ebruchez> sounds good to me
<John_Boyer> http://lists.w3.org/Archives/Public/public-forms/2009Feb/att-0084/11.2.g.xhtml
<prb> +1
<scribe> ACTION: John_Boyer to add test for using submission for validation purposes [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action05]
<trackbot> Sorry, couldn't find user - John_Boyer
http://lists.w3.org/Archives/Public/public-forms/2009Feb/0083.html
John: fine tuning of test suite
... probably lots of areas like this we could update in the
test suite
... we test for no nodes and produce error, then do validation
after that during submission...this is a test to ensure tests
happen in that order
... should we update or leave?
leave
is this a new test?
ok, then sure
adding
<prb> add
just didn't want to rock the boat if it's an existing test
<scribe> ACTION: John_Boyer to add test for sequencing of relevance pruning before validation [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action06]
<trackbot> Sorry, couldn't find user - John_Boyer
John: Paul, do you know if we add these tests to the core area, we have to add entries into xml results files, right?
Paul: yes
John: not clear to me how to wire them in to be run automatically
Paul: need to add new selenium driver for each test and add that to the main driver for that chapter
John: there's a real chance that folks don't get the processing order right
Paul: some impls may predate the spec nailing down this ordering
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0027.html
John: these tests need
modification
... have posted what's required
<John_Boyer> http://lists.w3.org/Archives/Public/public-forms/2009Mar/0027.html
John: looks like most would pass
the first test, and fail the second
... first is false positive, and second is false negative
... lazy authoring is when you don't have an instance
declared...but both tests have one
... the c.2 test has a ref="/car" and that node exists...
... nobody would report an error since the form is valid
... if you don't have an instance, then ref="/car" is not a
QName so we can't generate a node...so just need to remove the
instance so we can check QName conformance
... think I just removed the instance and cleaned stuff
up
... but implementors would need to be notified of this change
to retest
... should we change? think so
+1
<prb> +1
<scribe> ACTION: John_Boyer to update test suite to repair lazy authoring tests [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action07]
<trackbot> Sorry, couldn't find user - John_Boyer
John: automation is a bit
tricky
... since we can't initialize lazy authored nodes to a specific
value
... so we have to do a set and then test after the fact
... this aspect is good
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0045.html
John: test is claiming
requiredness from xsi:nil
... test author confused this with required MIP
... might be getting false positive passes here
... dont' have suggestion on how to rewrite this...will take a
bit of effort
Paul: what do you think it will take?
John: mostly to just rethink how
xsi:nil works etc
... this one is specifically testing xsi:nil for support
... not sure if anyone will pass this...option is to just
remove or rewrite it
<scribe> ACTION: wiecha to rewrite xsi:nil test [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action08]
<trackbot> Created ACTION-530 - Rewrite xsi:nil test [on Charles Wiecha - due 2009-04-01].
http://lists.w3.org/Archives/Public/public-forms/2009Mar/0046.html
John: title suggests data
mutations based on insert/delete...should just change to
include setvalue
... because of B11
... B9 is copy an attribute
<John_Boyer> http://lists.w3.org/Archives/Public/public-forms/2009Mar/0046.html
John: we have specific language requiring context attribute rather than nodeset to needs to change to reflect that
B11 says replace an attribute
attribute exists, and we're replacing with same name but diff value
best done by setvalue not insert
John: would like to update the
test to reflect this
... rather than delete followed by insert...two actions and a
rebuild vs. setvalue action
<scribe> ACTION: wiecha to update B9, B11, and B14 [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action09]
<trackbot> Created ACTION-531 - Update B9, B11, and B14 [on Charles Wiecha - due 2009-04-01].
<scribe> ACTION: John_Boyer to update B9, B11, and B14 tests [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action10]
<trackbot> Sorry, couldn't find user - John_Boyer
nick...please delete action 9 since john will do this instead
John: don't see an agenda item to
raise the issue of submission testing when replace="all"
... when we have tests looking for events coming off such a
submission, which is problematic
... most impls will delegate replace="all" to the browser so we
lose control
... if error, browser will report that
... if success, we're gone so how to report success?
... don't believe we have (or can have) passing impls for
this
... (checking...) not sure we actually have a test trying to do
this
Paul: think there is one, yes
John: we might have already
updated the test suite to correct for this
... it's still a problem for the spec but we might not be
testing for it right now
Paul: 11.4.b has replace="none"
John: also 11.4.a
... so they've been changed to do that
... so we have to worry just about the spec
11.4
John: language seems more
appropriate to replace instance and none
... in 11.2 possibility of amending first and last bullets to
avoid submit-done in replace="all" case
... objections?
... also submit-error
Erik: wondering whether in
javascript if you implement submission with replace="all" with
XHR could you still see these events?
... and server-side impls have more control over this
... so could perhaps also use these events
... so for those cases where you have control maybe we should
leave this
John: that's how we came to
remove link-error...implementation specific how they'd be
handled
... depending on whether the platform had control or not
... could add a note suggesting what some implementations could
do
Erik: native implementations could do this
John: rework bullets to add a note
Erik: or suggest that implementations may skip these events if unable to support them
John: might prefer implementation-specific behavior since this would not require tests or conforming impls
Erik: ok, in any event reasonable to alter this text
<scribe> ACTION: John_Boyer to update section 11.2 to suggest implementation specific behavior for submit-done or error [recorded in http://www.w3.org/2009/03/25-forms-minutes.html#action11]
<trackbot> Sorry, couldn't find user - John_Boyer
<John_Boyer> above action is specific to replace all case
John: all, please focus on
discussions related to test suite so we can close out the
implementation reports
... at least one more coming from the ubiquity side
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/should/issue list should/ Succeeded: s/suggesting/requiring/ Found Scribe: Charlie Wiecha Found ScribeNick: wiecha Default Present: wiecha, John_Boyer, Nick_van_den_Bleeken, ebruchez, prb Present: wiecha John_Boyer Nick_van_den_Bleeken ebruchez prb Regrets: Leigh Uli Agenda: http://lists.w3.org/Archives/Public/public-forms/2009Mar/0053.html Got date from IRC log name: 25 Mar 2009 Guessing minutes URL: http://www.w3.org/2009/03/25-forms-minutes.html People with action items: john_boyer nick wiecha WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]