Steven Pemberton, CWI/W3C (chair)
Alain Couthures, AgenceXML
Leigh Klotz, Xerox (minutes)
Philip Fennell, MarkLogic
Nick van den Bleeken, Inventive Designers
Erik Bruchez, Orbeon
John Boyer, IBM
Uli Lissé, DreamLabs
Steven Pemberton: I've removed
action items done by others, or with nobody.
Steven Pemberton: Please look at them
and update yours so we can focus on a couple next week.
Leigh Klotz: Does this need
action?
Steven Pemberton: It's sponsored by
W3C and others.
Steven Pemberton: Perhaps they sent it
to all groups.
Leigh Klotz: Kurt might be interested,
but he's not here today.
Steven Pemberton: Stricter error recovery, no absolute units except in print, font family names, changes in XHTML and whitespace. Do we feel we want to review this? No? CSS 2.1 has been so long, and nobody's been able to reference 2.1. We do want it to be a standard. I'll take a look at it but I don't think we need to review it.
Latest: http://lists.w3.org/Archives/Public/public-forms/2010Dec/0041.html http://lists.w3.org/Archives/Public/www-forms/2010Dec/0016.html
John Boyer: If there is a select1
item whose value element produces an empty string, we treat it as
matching the empty string in the data. In our implementation it's
hard to tell if you have a minimal appearance, however with
appearance=full it's easy to see that. The testing I
did, I made sure we weren't getting defaulting behavior, so I made
the empty item in the middle of a list to make sure it didn't
default to the first or last. When I selected an item in one
select1 it would change the other to the middle; it's clear that it
was working with empty values.
Steven Pemberton: I'm assuming we all
agree this is reasonable behavior. It seems like setting a value is
just as valid as anything else.
John Boyer: For the select1 case, it
seems like it would be possible to have a data value empty. If
separately required MIP or a schema forbids it, that's a form
design issue.
Steven Pemberton: There's the weird
case of select.
John Boyer: It's easier; the user can
go back and pick something.
Steven Pemberton: What do we say about
initialization? Does the empty item get pre-selected?
John Boyer: That's what happens in my
implementation. There's no way to tell if it's the empty data
template or data being presented previously filled out.
Steven Pemberton: Just checking. So
this is an erratum, or a clarification at least.
Leigh Klotz: I think it's an
erratum.
John Boyer: It's hard to tell if it's
an erratum or a clarification.
Steven Pemberton: But if we're
changing the meaning.
John Boyer: Either way, it's an
erratum.
Steven Pemberton: It sounds like
you've worked quite hard on this, John. Do you want to do the
initial draft?
John Boyer: Sure.
Steven Pemberton: If anyone's against
it, please take up now.
Resolution 2011-01-5.1: We accept the erratum that empty string as matching an empty item in select1 in XForms 1.1.
ACTION-1760 John Boyer to produce the erratum that empty string as matching an empty item in select1 in XForms 1.1.
Steven Pemberton: This looks like a typo we found converting to wikispec. It needs to be marked as an error in the spec itself.
ACTION-1761 Add two wikispec errata to the eratta document
Leigh Klotz: JavaScript compilation
is fast these days. This gives us XQuery 1.0 and hence XPath 2.0 in
the browser in JavaScript for XSLTForms and Ubiquity. So this makes
XPath 2.0 a reality.
Erik Bruchez: And Mike Kay has
announced compiling Saxon with XSLT into JavaScript.
Steven Pemberton: So including XPath
2.0 in XForms gets a higher priority.
Erik Bruchez: Is Alain
interested?
Alain Couthures: It's interesting and
it's already done. I'm also looking at XPath 2.0 libraries, for
scripts and XQuery Update.
Nick van: Interesting, they let you
update the dom http://www.xqib.org/js/SetStyle.html
Leigh Klotz: This will bring attention
to XML in the browser and we may be able to leverage that with
demos and announcements.
Steven Pemberton: I don't recall
what we decided about XPath 2.0. I thought we hadn't put it in and
had said XForms 2.0. Am I right?
John Boyer: No.
Erik Bruchez: http://www.w3.org/MarkUp/Forms/wiki/Category:XForms12
Erik Bruchez: It's in the list of
candidates. We haven't committed but my understanding is that
whichever modules get sponsored will move forward.
Steven Pemberton: Do we have two
implementations?
Nick van: Orbeon, Chiba/BetterForms,
modulo some corner cases.
Steven Pemberton: So creating the test
cases is teh work.
Leigh Klotz: And the data type
compatibility.
Nick van: And functions that return
new content.
John Boyer: And an attribute or some
other way to indicate from the get-go that XPath 2 is being
used.
Erik Bruchez: XSLT WG is doing a 3.0
rev of XPath and XQuery and XSLT.
Leigh Klotz: It's a rename of
2.1.
Erik Bruchez: Yes, it's not as major a
change as 1.0 to 2.0. But they have new things like first-class
functions.
Leigh Klotz: Didn't we ask for
first-class functions and they said no?
Erik Bruchez: Not quite. But we can
check. It's good for custom functions.
Leigh Klotz: Which is what Alain
mentioned.
Steven Pemberton: Who is the main
champion for XPath 2.0?
Nick van: That's me. I already
generated the functions in XPath 2.0 syntax; there are some to-do
items I'm picking up.
Steven Pemberton: We can plan a
meeting around XPath 2.0.
Nick van: The problem is I'll be on a
week's vacation in two weeks, so after that.
Alain Couthures: I've already
implemented it in XSLTForms with the Wikipedia JSON example and
I've written the paper.
Leigh Klotz: The message lists a lot
of issues: vCard, CSV, XML namespace rewriting, etc.. When will the
paper be published?
Alain Couthures: At XML Prague. I'm
submitting the abstract.
Steven Pemberton: Will we have to wait
or can we get an early version?
Alain Couthures: I'll have to discuss
it with them.
Leigh Klotz: Kurt replied under the
misleading subject "Re: Agenda for 2011-01-04".
Steven Pemberton: Alain, what is the
aim?
Alain Couthures: It's about XML and
HTML on the same page and how to deal with it, what parsers should
do (wrong or not), with and without namespaces, etc. It's a
difficult thread to follow. People feel the situation whether they
should use content types with text/html etc. I don't see very
clearly what progress there is yet.
Nick van: They are still trying to
figure out what they're trying to do first and are trying to
brainstorm. Now they're trying to find use cases; they've
identified a couple and are arguing about whether they're good.
People are brining in new information and harvesting use
cases.
Steven Pemberton: This group does have
a telephone call.
Nick van: They've had two calls.
Erik Bruchez: It looks like the
changes being discussed are mostly on the XML side. We had a
limited call a few weeks ago with Kurt and Leigh on this topic. One
question we had was embedding XML in HTML. Based on what I've seen,
it seems there is a major problem. If you want to have an HTML5
with XML islands, it's going to work in some cases, but in other
cases, the parser will change the layout of your XML document. If
it were possible to embed XML in HTML with structure preserved,
that would be good. The MathML have some issues. I don't know if
they will get to a solution. The point where it boils is that "it's
always been this way."
Nick van: They made it worse in HTML5.
It's now already in two browsers. There are more tags closed or
inserted.
Erik Bruchez: It seemed like there was
an idea that it had been there forever; there was a comment in
webkit that it's getting worse. So it's hard to understand what
changes are happening or are possible. It's good some of the XML
folks are paying attention.
Nick van: Now there are some
constructs breaking of "collateral damage" for recovery rules. It
affects us mostly with HTML in a label. If we use a <p>, then
<b> will break out of the </label>.
Steven Pemberton: Why?
Erik Bruchez: label is an inline
element.
Steven Pemberton: Our word label
clashes with their word label?
Nick van: label inside a p, for
example. You can't nest p elements. It automatically extracts
everything and closes the elements.
Erik Bruchez: One issue with MathML
was that if you had elements that looked like HTML inside the
MathML they got moved around. It's a general issue with
vocabularies: you can't do it reliably; you can't allow nesting in
certain cases.
Nick van: A table with a tr will
generate a tbody, to make CSS selectors work.
Erik Bruchez: It's a shame. Moving
around in the DOM has been there forever, but it's gotten worse as
you said by being made clearer. If you use the XHTML serialization
you don't have those issues. If you're telling people that if they
want to use MathML or XForms or something new cool and with
vocabularies and JavaScript they need to generate compliant XHTML,
it's a huge barrier to entry. It would be good if you could use the
HTML serialization.
Nick van: You have to set the
content-type to application/xml, because otherwise it will think
it's text/html.
Erik Bruchez: It's not as well
supported as the HTML parser. Firefox can't do progressive
rendering in the XML parser. It's supposed to happen soon. There's
a good case to be made for trying to understand if the things that
appear to be missing in HTML can be fixed. If really it's
inevitable that the custom markup people can't be satisfied and
everyone is convinced that the course cannot be changed, perhaps
something can be done.
Leigh Klotz: So the problem doesn't
happen with the application/xhtml+xml parser?
Nick van: Almost everything should
work but you have to specify the content-type.
Leigh Klotz: It may be that content
type starts working; they've been pushing this for years, I think
maybe Anne van Kesteren has talked about it for years. So maybe
they will get the content-type stuff to work and be widely
adopted.
Steven Pemberton: So that could be a
success.
Steven Pemberton: So Leigh asked if
we should have representation on this task force. Are we happy just
being on the mailing list? Do we want teleconference
representation?
Nick van: I'm watching it; Alain and
Kurt are participating.
Steven Pemberton: Do we want to be on
the phone call?
Leigh Klotz: Do we have a position
statement?
Nick van: We can wait until they are
done with use cases?
Leigh Klotz: Can we send in our use
cases?
Nick van: The most important want is
embedding XML in HTML and they list it. They also have XML
toolchain for XHTML5 documents.
Steven Pemberton: So we're happy for
now but we're happy not to be involved yet.
Nick van: Raman is also a part of the
task force; he's not on the call.
Leigh Klotz: Can we discuss XPath
2.0 next week?
Steven Pemberton: Nick said he can't
do it next week.
Leigh Klotz: Can we just remidn
ourselves of the issues?
Steven Pemberton: Sure.
Nick van: I'll try to have a list of
issues but I'm not sure I can.