See also: IRC log
<inserted> Scribenick: klotz
<scribe> Scribe: Leigh
John: Erik has asked about tests
that we may have nobody pass.
... Part of that is conformance triage.
... XForms 1.2
... For the latter half of the day, start on XForms for HTML
assertions
Erik: I won't be here this afternoon.
<wellsk> Nick, can you find out Joern meant with tests 5.5.2.a and 5.5.3.a passing? those are not real test cases
John: We got an idea of what
we're not doing (on Monday) but we need an idea of what we are
doing.
... Test suite? New report? 1.2 Mod?
Charlie: Let's look quickly at what's important from yesterday.
John: So, XForms 1.1
testing.
... Chapter 4 had some surprising no-passes.
Charlie: If there are ones people can sign up to implement...
<John_Boyer> http://lists.w3.org/Archives/Public/public-forms/2009Feb/0031.html
Keith: I made the test case changes to the Ubiquity tests but haven't tested them yet.
John: So we might get a few more
passes.
... 3.3.4.b more bind element examples
Erik: What should the first one do? There's no nodeset on the bind.
Nick: It is then the default context node, the root.
John: You should get an exception then as it has child elements.
Erik: Binding exception; that's new in 1.1
Keith: This was sparked by controversy on the list
John: The controversy was the second one, a bind with no nodeset, for inner binds.
<ebruchez> 8.1.1 "Element nodes: If element child nodes are present, then an xforms-binding-exception occurs. "
Leigh: So you should get an error on this whole page because of the first test, which has the wrong label?
Erik: So what does bind with no nodeset mean?
Nick: It's the initial context.
Erik: So bind_002 should be "avg" not "/car"
<Steven> i/scribe: Leigh/scribenick: klotz
Nick: There's no way to see both values at the same time.
Leigh: Should we look at the spec to see what this was trying to test?
John: It's testing the in-scope
evaluation context node as default for nodeset.
... [fixes example]
Erik: The /prices/price operates relative to the av element so we need
John: ..
Erik: We've lost the first bind
Leigh: That could be a different test.
John: The inner bind has no nodeset.
Leigh: OK.
John: 4.5.2.a xforms-compute-exception
Erik: The event handler is on the model; let me check the spec.
John: 4.5.2.a is a copy-paste error.
Erik: compute is dispatched to the model; there was a different test that was wrong.
John: Why is this a link exception?
Erik: The compute exception is always dispatched to the model; the bind exception is dispatched to the element.
John: It has to be changed to compute exception.
Erik: I think that's not true; it bubbles and stops processing after setting the instance.
Leigh: I think our exception handling is way over engineered for what value it provides.
John: It's under engineered; it just says "processing stops"
Erik: You can't recover from this one so it makes no sense.
John: We should say that the processor is required to dispatch the event but we can't say you can then use XForms markup.
Erik: It should disable xforms after that; the events bubble but no xforms; you could use script for example. We can't require the xforms engine to do anything.
Uli: We say processing stops, but I don't know what that means. What happens to outer events? Is it cut off?
Erik: It's implementation dependent.
John: It does capture, bubble, and then done. Anything after that nobody can have expectation.
Erik: You can set a global flag.
Uli: So how useful is such an exception if you can't handle it? It's the opposite of Java (error/exception)
John: We've changed version; we'll have to change this and some binding tests to "may also see the expression."
Erik: It's more correct.
... Will Ubiquity show it? Too early to say.
John: Too early ot say.
Erik: It depends on the processor.
John: Firefox should now pass this test.
Leigh: What generates it?
John: xf:output value="es@#invalidXpath"
Leigh: I think we've way overspecified what happens with authoring errors.
Erik: We should say that no XForms processing can be done after.
John: The spec says processing halt; the test assumes it doesn't.
Erik: I had never understood.
Uli: So it's useless to attach a message.
John: It says that the default action is to halt processing.
Erik: ...
John: We halt processing after.
Erik: I don't agree. We had this test where we through the exception during model initialization and the XForms processor could ignore any xforms actions that should be run.
<wellsk> FYI: test 4.5.a.xhmtl passes on FF3 when the link is changed to xforms-compute-exception
Erik: It's an edge case, not very useful. There's no way to say what ought to happen with half-initialized models and controls.
Uli: What's the user's expectation? The author? you want to be notified.
Leigh: I think it's a debugger issue, not an xforms issue.
Erik: Look at a javascript error, for example.
Uli: ...
John: This is a design tool mechanism; an xforms processor is required to dispatch the exception but other than dispatching the event and letting it bubble, our processing (actions, ui) is not to be expected. The XForms runtime is reporting an error.
Erik: That's for exceptions;
errors are recoverable. So we shouldn't spend too much time on
it. The spec should have a note saying that the minimum support
is necessary.
... Stop processing once default action is reached.
Charlie: Do you have to go to the default action? Can you dispatch and quit?
Erik: The default action is what stops processing. You could simulate it.
Charlie: It could terminate before bubble captures.
John: I've gone through once before and made sure all exceptions halt processing.
Uli: This says the default action happens after capture/bubble. I've learned it's not necessarily so, in discussions with Mark Birbeck.
Steven: XML Events doesn't define
it; it's DOM events.
... The DOM 2 spec is vague; as I understand it, it has to
happen after capture and bubble.
John: None are cancellable.
Leigh: can we just add a note saying what we "default action halts processing" means with respect to DOM events?
John: I think what we need to say
is that the hooked actions won't necessarily run either with
exceptions that happen early in model construction.
... We can make a note in each exception saying implementations
need not execute action handlers for each event (nor ui
refresh)
... This test 4.5.2.a is challenging
Erik: We can take out the second part. Just test to see that the exception is thrown.
Charlie: A script handler
John: Which is implementation dependent.
Leigh: Why not take out a "You may seen an output" and take out the Forms markup that runs after the exception. Leave in only the stuff that all implementations should show and just print out "processing should halt" and test the negative: put in markup that ought not happen to show processing has halted.
John: We did that once. Was it link exception?
Nick: 4.5.4.a
John: So all the exception should do like 4.5.4.a and say "You shouldn't see this" and say processor-dependent mechanism is necessary to test.
Leigh: Sounds good. If the xf:message in there isn't causing problems but won't show up necessarily, I'd say remove it, but I can live with it if it's not causing problems.
Erik: It's confusing because it's not required.
John: The xf:output inside xf:message may or may not work in the link exception.
Charlie: It's in the bubble-capture so it should work.
John: Since the link exception is
in the instance you might not have an evaluation context,
though it could be the default instance; we just don't say the
order.
... So should I amend all the tests now?
<scribe> ACTION: John Boyer, for each exception event, note that implementations are not required to run XForms action handlers, and user interface update should not occur. [recorded in http://www.w3.org/2009/02/11-forms-minutes.html#action01]
<trackbot> Sorry, amibiguous username (more than one match) - John
<trackbot> Try using a different identifier, such as family name or username (eg. jkugelma, jboyer)
<scribe> ACTION: jboyer, for each exception event, note that implementations are not required to run XForms action handlers, and user interface update must not occur. [recorded in http://www.w3.org/2009/02/11-forms-minutes.html#action02]
<trackbot> Sorry, couldn't find user - jboyer,
John: In 4.5.4.a we switched to using xhtml:p instead of xf:group because these exceptions halt processing.
<scribe> ACTION: John Boyer to update exception test cases to make no use of xforms processor after exception and indicate fatal error in prose. [recorded in http://www.w3.org/2009/02/11-forms-minutes.html#action03]
<trackbot> Sorry, amibiguous username (more than one match) - John
<trackbot> Try using a different identifier, such as family name or username (eg. jkugelma, jboyer)
Erik: ... xsi:type ...
Nick: There is a test using id
type as xsi:type
... There are 3 tests: name id, bind, and xsi:type
John: I have a preference for
things in our namespace.
... Do we need separate tests? I don't want to get rid of
xsi:type, but it looks like many fail.
Erik: We don't pass lots of type tests, just not via xsi:type.
Leigh: Is there a strong reason you don't implement xsi:type now?
Erik: I'd have to look.
John: It's a bit of a performance issue; we create pseudo-mips.
Erik: I could check at control binding.
John: It also affects validation.
Erik: Our validation process applies a schema and the goes through the binds. We are lacking the third one looking for xsi:type.
Leigh: So it's not more deep problems; it's just work.
Erik: Right.
John: What about the chapter 5
tests?
... So 5.1.a-d uses xsi:type. We need pairs with
bind/@type
... Or change these to use type mip and make one test for
xsi:type
http://xformstest.org/2009/02/f2f/tests now has 5.1.a
<John_Boyer> I added/uploaded a new test 5.1.e that tests datatype validation using xsi:type (in prep for converting the other tests to using the XForms type MIP)
http://xformstest.org/2009/02/f2f/tests now has 5.1.*
http://xformstest.org/2009/02/f2f/tests now has 5.2.1.*
http://xformstest.org/2009/02/f2f/tests now has 5.2.6.a
http://xformstest.org/2009/02/f2f/tests 5.2.1.a has changed the invalid anyURI value to "% 6 7" from " Bogus Data "
<Steven> Seen this XForms editor before? http://bpt.hpi.uni-potsdam.de/Oryx/XForms
s/5\.\2\,1\.a/5.1.a/
<John_Boyer> http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt04/4.5/4.5.3/4.5.3.a.xhtml
John: Can we test it in FormsPlayer?
Steven: Yes.
John: It notices and puts up an error message but doesn't run the action, so it seems possible but it's technically not passing.
Leigh: And show=replace
Steven: It shows it in a different place
Leigh: Why not remove the feature and let authors use submission/@replace='none' for testing beforehand and let authors who care rely on GET being idempotent?
John: That doesn't implement it properly because it does two gets.
Leigh: No, remove the feature and let authors do it themselves if they care.
John: It's a 1.0 feature
Erik: The world has changed since then and there are non-native implementations.
Leigh: It's possible to do as an author in HTML using XHR and two gets.
Erik: It's not quite correct because it might fail the second time.
Leigh: Yes, that's the Wile E. Coyote problem.
Erik: right.
... So it optional good enough?
John: Optional means one implementor.
Erik: What's weaker than optional?
John: Nothing.
... We did not in status distinguish between recommended and
optional. Maybe in the future.
Erik: So we can't go to rec.
Leigh: Unless we take it out.
John: We won't have a formal objection then.
Erik: Mark could make it pass.
John: But he's not working on
it.
... So what do you think about removing xforms-link-error? It
is a 1.0 feature but nobody is counting on it.
Raman: Remove it and see who screams.
<scribe> ACTION: John Boyer to remove xforms-link-error from 1.1 on the basis that it is not implemented. [recorded in http://www.w3.org/2009/02/11-forms-minutes.html#action04]
<trackbot> Sorry, amibiguous username (more than one match) - John
<trackbot> Try using a different identifier, such as family name or username (eg. jkugelma, jboyer)
http://inwap.com/pdp10/pclsr.txt
http://www.panda.com/tops-20/pclsr-memo.txt
http://lambda-the-ultimate.org/node/463 also on PCLSR
<Steven> scribe: steven
Charlie: It would be fun to experiment in Ubiquity with different frameworks in addition to Dojo
John: People outside of this room?
Leigh: We need to see the current
stuff working first
... I think I can get involved a bit further along, but not
yet
Charlie: Is there anyone who wants to help with Ubiquity
Steven: Not out of the
question
... now that Jack Jansen is adding SMIL to Ubiquity
Nick: Not on company time, but maybe in spare time
[John demos Ubiquity]
[with outputs with images in the labels]
[Group debugs example :-) ]
John: How much further along do we need to be?
Leigh: There are things that don't work that stop me from trying it out sufficiently
John: Couldn't you just fix it? It's open source
Leigh: Not enough resources
... I would like to see custom controls being possible
Steven: What's the process of joining
John: Join the contributors' group
<John_Boyer> http://groups.google.com/group/ubiquity-xforms-eng
John: There is more information
on that page
... including licensing
Steven: Install SVN?
John: Yes
Raman: There is a web interface
[John explains mechanics of Google code]
[Leigh tries following the instructions]
<wellsk> try this: http://code.google.com/p/ubiquity-xforms/source/browse/trunk/samples/purchaseOrder/purchaseOrder.html
<wellsk> oops: this one http://ubiquity-xforms.googlecode.com/svn/trunk/samples/purchaseOrder/purchaseOrder.html
<John_Boyer> http://www.w3.org/MarkUp/Forms/wiki/XForms_1.2_Modularization
John: We had a discussion Monday,
but I'm not sure what the resolution of that was
... is there a higher-level view? How do we move forward?
Leigh: Are we doing the modularization or not?
John: I think not, not at the
spec level
... there is a modularization at the script level
... We published XForms for HTML in December
Raman: What's stopping us from going to last call?
John: Last call in June would be
good
... taking us to CR
Steven: Who will implement?
John: Ubiquity
... success will be over different browsers
... since getting it to work in different browsers is a lot of
work
... it takes a lot of work to get through the rec track
process,
... another test suite on top of the current one
Leigh: We should write some samples
Charlie: There are some in the Ubiquity project
[John shows an example]
Leigh: We don't want to do full
modularization, but we are happy if it gets done for us by
Ubiquity.
... The first example we should look for is an instance
module
Charlie: Functionally that works today, but the modules are not available yet
Raman: Show how to do the Google suggest equivalent
Leigh: Pick an Ajax library, and look to see what it means for the instance module to work with it
Charlie: I am about to implement the data island module
Raman: You can sell it to people who like the model, and to people who need controls
John: Do people bind controls to JSON objects now?
Charlie: ...
... there are data providers that give notifications
... The repeat module is clearly missing elsewhere, so is a
strength for us
John: So are we done?
Uli: Well, there's XForms 2.0
Leigh: We still have to deliver
on the charter
... but modularization is not directly serving anyone
... we should be open to take the modules from outside, but we
don't have to do it ourselves
<John_Boyer> work items going forward, then, are
<John_Boyer> 1) advancing xforms for html (examples, assertions, test suite),
<John_Boyer> 2) some new xforms 1.2 features to support xforms for html, such as context everywhere and json submissions
<John_Boyer> 3) start analyzing ajax programming and how parts or all of xforms might fit in
<John_Boyer> 4) XForms 2.0
<John_Boyer> 5) Finish XForms 1.1
Leigh: Why do we have to do Basic?
Steven: We can't leave it; either a note or take it to rec
Leigh: 1.1 has subsumed basic in a profile
Raman: Just do it, it is the same amount of work
Leigh: We should add something to the status that points to 1.1 basic profile
John: I favour a note, as being less work
opic: Future meetings
Steven: Amsterdam or London?
Leigh: Amsterdam is easier to argue for
Charlie: I'd like to go somewhere new
Steven: Well, we don't have to decide now. We can just say Amsterdam/London
Uli: London will be more expensive
Raman: But in London Mark will come
Steven: He may well come to
Amsterdam
... it's not too far
... And then there's TPAC
... Dates for Europe?
Nick: 8-10 June
... 4 June is the vday
John: London for preference?
Uli: Yes
Leigh: Slightly harder for me then
John: When and where is TPAC?
Uli: San Francisco area, 2-6
November
... 2-6 November
TPAC2009, Santa Clara Marriott, Santa Clara, CA, USA (in conjunction with AC Meeting)
<unl> http://www.w3.org/Consortium/meetings
http://www.w3.org/Member/Eventscal
http://www.w3.org/Member/#foryourcal
<nick> http://www.w3.org/Consortium/meetings
<klotz> http://letmegooglethatforyou.com/?q=w3c+technical+plenary+2009
http://www.w3.org/2002/09/TPOverview.html
EOM?
[ADJOURN]
<nick> 11.10.a
<nick> 11.4.b Corrected instance reference
<nick> 5.1.a Changed xsi:type to MIPs
<nick> 5.1.b Changed xsi:type to MIPs
<nick> 5.1.c Changed xsi:type to MIPs
<nick> 5.1.d Changed xsi:type to MIPs
<nick> 5.2.1.a Changed xsi:type to MIPs
<nick> 5.2.1.b Changed xsi:type to MIPs
<nick> 5.2.1.c Changed xsi:type to MIPs
<nick> Updated those action items
<wellsk> got them
<John_Boyer> also previously checked in new 5.1.e
This is scribe.perl Revision: 1.133 of Date: 2008/01/18 18:48:51 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/triange/triage/ Succeeded: i/John: Erik has asked about tests /Topic: Testing 1.1 FAILED: i/scribe: Leigh/scribenick: klotz Succeeded: i/Scribe: Leigh/Scribenick: klotz Succeeded: s/nad/and/ Succeeded: s/iwth/with/ FAILED: s/5\.\2\,1\.a/5.1.a/ Succeeded: s/tu/ut/ Succeeded: s/than/in addition to/ Succeeded: s/Noton/Not on/ Succeeded: s/How/How / Succeeded: s/group/ group/ Succeeded: s/AM/Am/ Succeeded: s/AN/An/ Succeeded: s/STeven/Steven/G Found ScribeNick: klotz Found Scribe: Leigh Found Scribe: steven Inferring ScribeNick: Steven Scribes: Leigh, steven ScribeNicks: klotz, Steven Present: Raman Hubble Uli Leigh Charlie John Erik Nick Steven Keith(remote) Got date from IRC log name: 11 Feb 2009 Guessing minutes URL: http://www.w3.org/2009/02/11-forms-minutes.html People with action items: are boyer each event exception for implementations jboyer john not note required that[End of scribe.perl diagnostic output]