John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Paul Butcher, WebBackplane
Steven Pemberton, CWI/W3C
Uli Lissé, DreamLabs
Charlie Wiecha, IBM
Erik Bruchez, Orbeon [joined late]
http://lists.w3.org/Archives/Public/public-forms/2009Feb/0061.html
John Boyer: It's hard to do
lazy-author parsing with just an output/@value.
John Boyer: Uli suggested we create
default instance data for cases where we don't have any UI binding.
My read is that a value is an expression. We limit ref in lazy
authoring to non-colonized refs. It might be more difficult to
parse XPath expressions looking for NCNames. I think we just added
an instance to the test case, but does anybody want to work on
it?
Leigh Klotz: One option is that we
could combine Uli's proposal of an empty instnace, and then
re-define the parsing in terms of insert events.
John Boyer: That's an interesting
refinement. Even if you define it in terms of an insert action, I
don't know that that solves the issue of what to insert based on
the value expression.
Leigh Klotz: At least you wouldn't
have no instance, so you would have a context, so output/@value
would work.
John Boyer: We could you get a default
empty instnace. Then if we find any refs, we add them in. I don't
know that we have to do them via an insert. That could be
disconcerting. I don't know that we need people hooking into
it.
Leigh Klotz: Starting with an empty
instance and adding to it seems more explainable to me.
John Boyer: Is it enough that we fix
the test case for now or should we fix XForms 1.1?
Leigh Klotz: I think it's only
important for XForms for HTML, and it's not thought out enough for
XForms 1.1
John Boyer: OK, so it's for XForm 1.2,
and we fix the test case to have an empty instance.
Resolution 2009-02-25.1: For http://lists.w3.org/Archives/Public/public-forms/2009Feb/0056.html we fix the test case and defer possible changes to processing of lazy-authoring empty instances to XForms 1.2.
Is further response needed on xf:copy on select1 http://lists.w3.org/Archives/Public/www-forms/2009Feb/0019.html
John Boyer: Aaron observes that with
xf:select1, xf:value is different from xf:copy. That same
inconsistency doesn't happen with xf:select because the value case
in select matches a particular substring within a space-separated
list and does the logical equivalent of the copy matching, where it
only inserts or deletes content that matches what the copy or value
element indicates. But in the select1 case with a value element, we
destroy the entire content of the node referenced by the select1
and insert the textual content of the selected item, whereas in the
case of a select with xf:copy, when you select one item, whatever
gets deselected we delete and so if there is surrounding whitespace
or other elements within the node referenced by the select1, those
don't get deleted. I believe that we did that sort if intentionally
because, not that we care about supporting select1 referencing a
node with three element children and two are bogus and one
corresponds to item b, we want the item b deleted but not the other
cruft items. I don't think we were trying to support the case of
keeping the cruft around, but authoring convenience for simple
whitespace indenting. If you imagine a node with whitespace and the
matching node and more whitespace, the whitespace text nodes don't
disappear. Or, when we defined copy with insert and delete, it
wasn't clear we had a match because the additional whitespace nodes
were in the content.
Leigh Klotz: But Philip's use case is
the "non-matching cruft" case.
John Boyer: Yes, that's the select
case. We also have the "younger brother" problem with the select1;
it's whitespace. But the majority of the discussion was about
select.
Leigh Klotz: You can't tell whether "a
b c" is a space-separated list or a single value, whereas you can
tell definitely with elements.
John Boyer: We went for the
Procrustean replacement of the entire content, which is how
xf:input behaves. There's no way to bind to the internal content of
the select1 case.
Leigh Klotz: What if the author
associates the listItem type with it?
John Boyer: Even if not, you could try
to replace the value substring and delete the substring.
Leigh Klotz: Is that what Aaron's
asking us to do?
John Boyer: I think he's pointing out
the inconsistency and asking us what to do. I'd rather than copy
not behave inconsistently with select and select1, and recognize
that value does behave differently. We have special-requirement
listing structure and therefore the structure existed to do
something better in select, so we can do substring matching.
Leigh Klotz: So in answer to my
question, we could say select could bind to listItems, and if
select1 bound to listItem, it would replace the whole content; if
somehow you could bind select1 to listItems, you could get the
partial-replacement of internal text replacement, but we don't
specify select1 has special behavior when bound to listItems; it
simply acts like...
John Boyer: like setvalue. And in the
copy case, we aren't bound to simple content, and so we don't do
setvalue, we do insert and delete. That's kind of nice because we
have defined our UI controls in terms of our actions. There's no
magic.
Leigh Klotz: So do you want me to
answer?
Action 2009-02-25.1: Leigh Klotz to answer http://lists.w3.org/Archives/Public/www-forms/2009Feb/0019.html with the discussion from the minutes.
Handling unspecified and unsupported submission methods http://lists.w3.org/Archives/Public/public-forms/2009Feb/0082.html http://lists.w3.org/Archives/Public/public-forms/2009Feb/0062.html
John Boyer: So if we make
get the default, do we want it fixed sooner? So it's
for 1.2?
Steven Pemberton: OK.
Resolution 2009-02-25.2: We re-visit http://lists.w3.org/Archives/Public/public-forms/2009Feb/0082.html and http://lists.w3.org/Archives/Public/public-forms/2009Feb/0082.html for XForms 1.2.
Delete processing clarification http://lists.w3.org/Archives/Public/www-forms/2009Feb/0023.html
John Boyer: What happens if you
refer to a namespace node in XPath.
Paul Butcher: You can.
John Boyer: What happens if you delete
it.
Steven Pemberton: Whoa!
John Boyer: I think he suggests we
ignore it. We have some ignore cases in insert. We probably need
some in delete. We say you can't delete the document element, for
example. He says we should do the same for the root node. He also
points out these issues for insert and setvalue.
Leigh Klotz: We can't really say that
we prohibit non-DOM operations on the basis that we require DOM,
because we don't require DOM.
John Boyer: Vlad's point is that the
inherited namespace nodes are fictitious. You'd have to delete it
at the ancestor level, from a DOM perspective.
Nick van: That's not correct; if you
still have elements in the namespace, the namespace nodes will
still be there.
John Boyer: I think if you allowed
deletion of that namespace node the XPath data model would be
inconsistent.
Nick van: But after you normalize the
document, the attribute will be there again, in DOM.
John Boyer: He argues that it it's an
invalid operation, and that namespace nodes are immutable.
Nick van: Isn't the XPath data model
immutable by definition?
John Boyer: I guess you're right;
there's no discussion of mutation.
Nick van: It's a snapshot...
John Boyer: ...of what's happening at
the time. We're defining operations on the XPath data model and as
such these are operations that make it not an XPath data
model.
Leigh Klotz: My point was that we
can't say "You can't do operations on nodes that aren't supported
by DOM." We have to be specific.
John Boyer: For what it's worth, the
things to prohibit so far are insert/delete/setvalue of namespace
node; insert/delete/setvalue of root nodes.
John Boyer: So we don't need
interoperability testing because these things aren't possible. We
should just say they are no-ops. Any objections?
Charlie Wiecha: Sounds good.
Resolution 2009-02-25.3: We accept implementation feedback for XForms 1.1 http://lists.w3.org/Archives/Public/www-forms/2009Feb/0023.html
Action 2009-02-25.2: John Boyer to amend XForms 1.1 in accordance with http://lists.w3.org/Archives/Public/www-forms/2009Feb/0023.html to change insert, delete, and setvalue to say that they ignore changes to namespace nodes and root nodes.
Steven Pemberton: They want to know
if you aren't going to attend TPAC in November in California, they
are considering cancelling it. So we are to discuss it.
Leigh Klotz: I think Erik and I can
make it...
John Boyer: It's reasonable to assume
I can make it.
Steven Pemberton: I plan to go.
Nick van: It will be difficult for me.
If we're talking about new features, that would be helpful.
Charlie Wiecha: It's important for me
for Backplane; I'll try to go.
Nick van: We are in favor of a cheaper
solution and moving hotels, and my company has asked me to say that
we are in favor. Moving from November to next year would be a big
gap then, from June to February.
Steven Pemberton: We could have a
virtual meeting in November.
John Boyer: We have two virtual days
already in November.
Steven Pemberton: So our reply is that
there's a good chance we will attend and that we think it is
valuable.
Nick van: Normally, until now the TPAC
have been in expensive hotels. We would prefer a not so expensive
hotel.
Steven Pemberton: There's no
requirement that you go to the hotel.
Nick van: But they do ask that you go
there for the rooms.
Resolution 2009-02-25.4: The Forms WG sees value in the November TPAC and many members plan to attend.
Steven Pemberton: Forms WG members
should attend the AC meeting as there will be discussion of XForms,
HTML5, and XHTML. There is always the possibility of joining by
phone, but they are planning breakout groups which makes remote
joining harder.
John Boyer: Charlie, you're going to
talk about AJAX library in the browser?
Charlie Wiecha: Yes.
John Boyer: I saw the XHTML2/XForms
discussion together.
Steven Pemberton: We do get grouped
together.
Charlie Wiecha: It's XML on the
client, all together.
John Boyer: And XForms for HTML?
Charlie Wiecha: I will bring it
up.
John Boyer: Erik, did you get CVS
access?
Erik Bruchez: Not yet.
Steven Pemberton: I requested it. They
may be really busy at the moment.
Erik Bruchez: It was xf:output with
label only. 3.2.2.h.
John Boyer: Would you fix others
similarly?
Erik Bruchez: I'll check.
John Boyer: Please send a note to
public-forms and cc www-forms when you check them in.
John Boyer: Nick?
Nick van: I started on the encoding
but haven't finished.
John Boyer: Will you be able to
indicate which tests have changed?
Nick van: There were about 12.
John Boyer: Thanks to Nick to help us by rolling up the test suite changes and notifying people.
Paul Butcher: I've found some
possibly erroneous tests I've found with Ubiquity and listed them
in the Ubiquity tracker.
John Boyer: That would be great.
Nick van: There are a number of
tests with ASCII encoding: 185 of them.
Leigh Klotz: I would count that as one
change myself. Plus the ones that we fixed at the F2F for other
reasons, to there were more.
John Boyer: This one Nick pointed
out sets up a handler for xforms-link-exception, which won't
happen.
John Boyer: I thought I was supposed
to get an action item to amend XForms 1.1 for link and version
exceptions to say that we don't expect the UI to be running, nor
action handlers to fire.
Leigh Klotz: The F2F actions are all
in the IRC log.
Nick van: I already processed the IRC
log.
Action 2009-02-25.3: John Boyer to amend XForms 1.1 for xforms-link- and xforms-version- exceptions to say that we don't expect the UI to be running, nor action handlers to fire, as discussed at the Feb 2009 / Google / Mountain View F2F
John Boyer: We don't need to test the event function in this way. We have another test. I think we should just eliminate this test. Anybody uncomfortable?
Resolution 2009-02-25.5: We drop test 7.11.2.b because there is another test of the event function and this test is erroneous.
Action 2009-02-25.4: Nick van den Bleeken to drop XForms 1.1 test 7.11.2.b
John Boyer: There is a test for xforms-link-error event dispatch for a load action. FormsPlayer does detect a bad URL for a load; formsPlayer will test you that it's unable to access it. Unlike other implementations, which simply delegate the load to the browser. The actual test suite test claims that the author could have a handler for xforms-link-error. formsPlayer does not seem to support the dispatch and handling of that event. So at the F2F we resolved to get rid of the feature. We figured the formsPlayer team wouldn't mind. formsPlayer can continue to report the problem to the user.
Action 2009-02-25.5: John Boyer to remove xforms-link-error from XForms 1.1
Action 2009-02-25.6: Nick van den Bleeken to remove xforms-link-error tests XForms 1.1
Action 2009-02-25.7: John Boyer to remove resource-uri example from event function in XForms 1.1.