John Boyer, IBM
Nick van den Bleeken, Inventive Designers
Philip Fennell, MarkLogic
Steven Pemberton, CWI/W3C (chair)
Leigh Klotz, Xerox (minutes)
Alain Couthuries, AgenceXML
Dan MacCreary, Dan MacCreary Associates
Erik Bruchez, Orbeon
Kurt Cagle, XMLToday
Steven Pemberton: Nick and I did this at the editorial workshop. I found some missing minutes and added them, and added the TOC of each.
Steven Pemberton: It's been
announced now.
John Boyer: The question is whether we
can become a community group, or whether that's appropriate.
Steven Pemberton: As it's already a
working group we can't move.
Leigh Klotz: Or we could seek
proposals for an XForms 2 there.
John Boyer: We can't do PER from a
community group.
Nick van: We copied XForms 1.1 to
XForms 1.2 in wiki markup and Steven re-arranged it to make it more
readable. I did some work on the XPath 2.0 part. Steven also
started work on @ref instead of @nodeset, for the resolution we did
a couple of months ago.
http://www.w3.org/MarkUp/Forms/specs/XForms1.2/modules/xpath20/index-all.html
Nick van: For the XPath 2.0 module,
it's a good time to have the group review.
John Boyer: How are the edits?
Steven Pemberton: A lot is similar to
what we already have.
Steven Pemberton: That's not yet done.
It was a productive two days. The editing is not yet finished, but
looking OK. There's stuf to be copied from the new wiki sections.
We made a good first version.
John Boyer: On the @ref part, how did
you do it? @ref is now offered by SNB, but we didn't offer that in
the @nodeset places.
Steven Pemberton: My current plan is
still to have SNB but the other also links to the same section
where the syntax is the same, and there's an extra rule in SNB to
take the first node.
John Boyer: So you still use SNB and
nodeset binding.
Steven Pemberton: Yes, it's a semantic
difference, but for multiple binding there is a deprecated @nodeset
attribute.
Steven Pemberton: We had a number of
questions left as notes in the spec.
John Boyer: I see it's 2.0.
Steven Pemberton: I thought we'd
decided to go to 2.0 because we're going to XPath 2.0.
John Boyer: I hadn't recalled that. I
thought XPath 2.0 was recommended.
Steven Pemberton: We could have a 1.2
being the same with just XPath 1.0.
Nick van:
http://www.w3.org/2007/10/htmldiff?doc1=http%3A%2F%2Fwww.w3.org%2FMarkUp%2FForms%2Fwiki%2FXForms_2.0_base&doc2=http%3A%2F%2Fwww.w3.org%2FMarkUp%2FForms%2Fwiki%2FXForms_2.0
John Boyer: We have XForms 1.2
Category on the wiki.
Steven Pemberton: That was before the
decision, but that was my memory.
John Boyer: I don't know when that
decision happened. XForms 1.2 Category includes XPath 2.0.
Nick van: In the current 2.0 spec it
can still be 1.2, but we did some edits to abstract away the
expression language so there is no normative reference to XPath.
The XPath 2.0 module brings in XPath 2.0.
John Boyer: How do you change
expression language?
Steven Pemberton: Only by
reference.
John Boyer: How can the processor
know?
Steven Pemberton: I guess the version
attribute.
John Boyer: Or it could be a separate
version for expression language.
Nick van: That's still open for more
expression languages.
John Boyer: As XForms 1.2 it's
important as I'd like to have features in my processor added, and
I'd like to get at those without having to do the expression
language.
Nick van: There is an xpath-version
attribute 1.0 or 2.0 on the model. If there's no attribute it's
1.0.
John Boyer: Doesn't the model element
itself have to list that element.
Nick van: Other modules can add
attributes to elements.
John Boyer: What's the default?
Nick van: I wasn't sure how we were
going to say how to combine the modules.
Leigh Klotz: So just like saying you
use XHTML 1.1, SVG 2.0 and XForms 1.1, you'd say you have XForms
2.0, SVG 2.0, and XPath 2.0?
Nick van: Someone has to say. If we
add more expression models, the processor needs to know.
John Boyer: It seems like you need an
XPath 1.0 module as well.
Nick van: It depends, perhaps not yet.
If you put the XPath 2.0 module in backward-compatibility mode it's
almost compatible with an XPath 1.0 implementation. We could have
an XPath 1.0 module.
John Boyer: One way of implementing
that module is using 2.0 in 1.0 mode, and another is to use a 1.0
processor.
Steven Pemberton: Is 2.0 in 1.0 mode
sufficiently different? You just need to explain how 1.0 works and
if we have that with the existing module? Maybe a section saying
backward compatible mode?
John Boyer: Then maybe it's the XPath
module and explains 1.0 and 2.0 modes?
Steven Pemberton: That might be
it.
John Boyer: In XPath 1.0 we have
non-ns function names.
Nick van: There are some subtle
differences. In 1.0 compat mode you can use sequences, but a pure
XPath 1.0 implementation does not. It's handy to use
sequences.
Leigh Klotz: John, is this a
hypothetical concern or a concern for your implementation?
John Boyer: I have two engines using
XPath 1.0.
Leigh Klotz: Do people import into
your processor?
John Boyer: Yes, we have
importers.
Leigh Klotz: So your concern is
someone writing a form under the guise of XPath 1.0 but it includes
sequences which are really 2.0 and your processor cannot run
it?
John Boyer: Right.
Steven Pemberton: What 1.0 expression
would have semantic differences.
Nick van: You can have an XPath 2.0
expression in XPath 1.0 compatibility mode with existing queries
being compatible with 1.0.
John Boyer: Some expressions would
stop working because of type problems with an XPath 2.0 processor,
but the 1.0 expressions would then work. So that's transitional.
Add features to the form and even some features, but still use
predominantly XPath 1.0 expressions.
Nick van: That's how we do it in our
products. We have a lot of customers using XSLT 1.0. We'd like the
spec to have an XForms 1.1 form work with an XPath 2.0
engine.
John Boyer: Then we have XPath 1.0,
XPath 2.0 with 1.0 compatibility, and true XPath 2.0. We're saying
that the True XPath 1.0 implementation may choose to use XPath 2.0
in compatibility mode. But that will not be able to do real XPath
1.0.
Leigh Klotz: That doesn't seem like a
realistic concern. There's XPath 2.0 implementations of XForms.
XPath 2.0.
John Boyer: I don't have a C or
JavaScript version of XPath 2.0 I can use.
Leigh Klotz: I'm not talking about new
implementations. Many existing implementations already use XPath
2.0 and aren't going to offer XPath 1.0 when the upgrade to XForms
2.0.
Nick van: Do we need XPath 1.0?
John Boyer: If it were XForms 1.2 yes,
it's a dot-release. You can still use most of what you already
know. I wasn't aware we weren't doing XForms 1.2. I thought we
agreed XForms 2.0 would go to XPath 2.0. But in the charter and the
Wiki I thought we were doing XForms 1.2. You could use XPath 2.0 if
you turn it on the markup.
Steven Pemberton: If we have XForms
2.0 with XPath 1.0 allowed, you'd get the new features and get
XForms 2.0 "basic"?
John Boyer: If we're going straight to
XForms 2.0 the default should be XPath 2.0. You'd have to ask for
XForms 1.0. The major version shift makes it different.
Steven Pemberton: There are questions
about how we identify versions.
Leigh Klotz: I don't see why you need
to specify it when pull XPath out into a module.
John Boyer: You'd have to specify
which XPath version you want.
Leigh Klotz: It doesn't have to be
XPath the way the spec is now.
Kurt Cagle: Would anyone write
XForms 2.0 with XPath 1.0?
Steven Pemberton: Yes, John doesn't
seem to want to write a new XPath engine.
John Boyer: Yes, and we need to keep
existing forms working.
Steven Pemberton: Don't existing forms
work in compatibility mode?
Nick van: In compatibility mode it
works. In 2.0 mode you may have type casting errros, saying the
arguments are not of the correct type. For example lt and gt can't
compare strings and numbers.
John Boyer: We have a problem layered
over top of XPath. For historical reasons we have no-ns qualified
XPath 1.0 mode functions. In XPath 2.0 you have to use NS
qualifications.
Leigh Klotz: John, can't you write an
upgrader?
John Boyer: That's about as much work
as putting in a new XPath engine, but ultimately yes.
Nick van: If you ask for XPath 1.0
implemented with XPath 1.0 or XPath 2.0 with compatibility mode,
then it's just one paragraph.
John Boyer: Yes.
Nick van: If you use XPath 2.0 stuff
in XPath 1.0 compatibility mode then it doesn't work in XPath 1.0
processors.
John Boyer: It is the most reasonable
standard protection one can get at this point to say it's legal to
use an XPath 1.0 processor.
Nick van: Then we need to change the
title to XPath Module.
Steven Pemberton: Yes, that's what
John said.
Nick van: I called it XPath 2.0
because XPath 3.0 is almost out.
John Boyer: To me the idea is that
XPath Module is out and it has text about 1.0 and 2.0. You can rev
it to include 1.0, 2.0, and 3.0 or maybe 2.0 and 3.0.
ACTION-1806 Nick van den Bleeken to change the title of the XPath 2.0 Module to XPath Module.
Steven Pemberton: Nick and I will
work on this until it's time for the vacation. We'll do this next
in the Bay Area.
Dan MacCreary: I want to acknowledge
how much work you've done.
Steven Pemberton: There's a lot of
detail. We think we can diff the versions as well to produce a diff
spec.
Dan MacCreary: Is there an easy way to
make sure the wiki changes are in?
Steven Pemberton: In a way we're all
editors now, but there's no objection to making agreed changes. It
isn't limited, just to WG members and we can make changes.
Dan MacCreary: Using the wiki
tool?
Steven Pemberton: Yes, that's a live
copy. You can move your section straight in.
Nick van: Yes. There's also and
XForms_2.0_base and don't touch that. That's the diff base.
Steven Pemberton: It's open for all to
edit. Just fix typos and it's recorded in the history page. Don't
edit the whole spec, just the sections, or make sure your browser
can handle the whole text.
John Boyer: should also make sure that
spec allows an xpath 1.0 to be implemented either with a real xpath
1.0 engine or with an xpath 2.0 engine with 1.0 compat mode
John Boyer: And add your name to the
editor list?
Steven Pemberton: Frankly we're all
editors, but if you make changes, by all means...
John Boyer: And just adding your name
doesn't count.
Leigh Klotz: Summer calendar?
Steven Pemberton: Yes, I'll do
that.