Steven Pemberton, W3C/CWI
Charlie Wiecha, IBM
John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Lars Opperman, Sun
Mark Birbeck, x-port.net (IRC)
Mark Seaborne, PicoForms
Nick van den Bleeken, Inventive Designers
Rafael Benito, SATEC
Roger Perez, SATEC
Sebastian Schnitzenbaumer, DreamLabs
Blake Jones, ViewPlus Technologies/DAISY
Summer Questionnaire: http://www.w3.org/2002/09/wbs/32219/xformssummer07/
Please fill in: Lars, Rogelio, Sebastian, Joern, Leigh
John Boyer: Leigh, are you away any
more in July and August?
Leigh Klotz: I am done with
vacation.
Attendance Questionnaire: http://www.w3.org/2002/09/wbs/34637/ftfsept2007/
John Boyer: I've got us past pub
rules and link checks. I saw you had to do a lot of work Steven
last time.
Steven Pemberton: Yes.
John Boyer: Do we have the entrance
criteria met?
Steven Pemberton: [reads] same as PR,
shows implementation experience that supports the changes.
John Boyer: How do we do that?
Steven Pemberton: What are the
changes?
John Boyer: 41 errata, some trivial
and not implementation changes. Our implementation can demonstrate
most of them. I'm particularly concerned about the version
attribute.
John Boyer: If you set
xforms=1.0 then existing processors won't do it.
Steven Pemberton: So if this is a
substantive change to 1.0 processors they should stop, that means
we have to run test suites.
John Boyer: Is that true only of
conformance level errata?
Steven Pemberton: Not for typos, as it
doesn't affect any existing browsers.
John Boyer: How did we do this for 1.0
SE? We didn't have an updated implementation report.
Steven Pemberton: Did we have
substantive changes?
John Boyer: We had at least one; but
we had conformance as being schema changes.
Steven Pemberton: What was that
change?
John Boyer: It was the addition of the
instance attribute to submission.
Steven Pemberton: In that case we
argued that it was left out by accident and we already had
implementations doing it; we they just accepted our word for it
without a test suite.
John Boyer: Here's the 1.0 TE in
editor's draft format:
http://www.w3.org/MarkUp/Forms/specs/XForms1.0.ThirdEdition/index-all.html
Steven Pemberton: Let me see what I
said then.
John Boyer: The three errata were E9
(id), and people already did that; E32F (switch inside repeat), the
same.
Leigh Klotz: And there was a note in
the spec asking for it anyway.
John Boyer: Then then was the version
attribute. It's obvious that it can be done; there's no mystery.
But we need an implementation or two to have actually done
it.
John Boyer: Are there any who have
done it? We're down Mark and Erik.
Leigh Klotz: Most implementations with
1.1 features now are 1.0 with some 1.1 features. What would you
like to happen in them if version=1.1?
John Boyer: The spec says that if it
is not 1.1 conformant then...
Leigh Klotz: But nobody is 1.1
conformant, or can be.
Steven Pemberton: http://lists.w3.org/Archives/Member/chairs/2005JulSep/0020.html PER request from Masayasu "There is a correction that affects conformance. Erratum E69a [3] adds "instance" to the list of attributes for "submission". The WG believes that without this correction, the intended use of instance replacement is typically not achievable in practice. We have coordinated with implementors on this issue, and actually this is what implementations already do. So this correction was added to align specification with implementations."
John Boyer: So what about Leigh's
question? Since we're not at PR yet it can't be done.
Leigh Klotz: So version missing is
already handled by 1.0; version=1.0 isn't but could be
easily. But what do processors with 1.0 but some 1.1 features do
with version=1.1?
John Boyer: How about we make it a
recommendation for 1.0?
Sebastian Schnitzenbaumer: A should
instead of a must. That would make it an informative
addition.
John Boyer: It would be normative, but
optional.
Sebastian Schnitzenbaumer: How does it
correlate with the test suite? If it's normative?
Steven Pemberton: I believe I heard
that every feature should be exercised.
Sebastian Schnitzenbaumer: I thought
that normative "should" still had to be tested for
interoperability.
Steven Pemberton: Yes, two
interoperable implementations for "should." But on the "may"
level...
Sebastian Schnitzenbaumer: Should we
move this to the "should" level?
Steven Pemberton: No, we can just say
"For future compatibility, we strong recommend that XForms 1.0 user
agents recognize the version attribute and act on it."
John Boyer: That's a should.
Sebastian Schnitzenbaumer: Then we can
say it's an informative paragraph.
Steven Pemberton: That's equivalent to
a may because existing implementations will still be
valid. We'd like it to be more than may. "We would
really like you to do it."
John Boyer: We can put in an
informative note saying we strong encourage it and say that
Steven Pemberton: Make it a "may" and
then say in a note "This attribute will be added to future versions
of XForms and we strongly encourage it"
John Boyer: Then I can remove that
erratum from the status section, and he others are already
implemented.
Resolution 2007-07-18.1: We change the
version attribute erratum to may and include a note
saying that it's for forward compatibility and implementors are
encouraged to add it. We remove it from the list of changes that
require test suite changes.
Action 2007-07-18.1: John Boyer to implement Resolution 2007-07-18.1 .
John Boyer: Do you need to be here?
Can you request it before vacation?
Steven Pemberton: Yes, I can request
it for the week I get back.
John Boyer: Then I send the
transition request.
Steven Pemberton: Masayasu did both
last time.
Steven Pemberton: And then the
dependencies, but it looks like I have to be there anyway.
John Boyer: The HTML WG has no
dependency on XForms 1.0, just XForms 1.2. And perhaps the Schema
WG chair because of the types.
Steven Pemberton: They don't have to
be invited; they just have to be informed, and turn up if they find
problems.
John Boyer: So 7 days prior, etc. Can
you send the request out this week?
Steven Pemberton: Yes. We'll be
waiting on Steve Bratt.
John Boyer: And the teleconference
bridge.
Steven Pemberton: That's easy.
John Boyer: Do I send the transition
request before you do that?
Steven Pemberton: Yes, then I can
point to the request.
John Boyer: Segue to ... Is there
anyone who objects to our seeking the transition to transition to
PR?
Steven Pemberton: Does anybody object
to taking XForms 1.0 Third Edition to PER?
Steven Pemberton: Next F2F?
Sebastian Schnitzenbaumer: http://www.w3.org/MarkUp/Forms/Group
redirects to http://www.w3.org/MarkUp/Forms/Group
and the second item on the F2F gets you a 404.
John Boyer: I made a cut and paste
typo. I added that link yesterday.
Steven Pemberton: The week of the
tenth.
John Boyer: I just changed the 06 to
09 and didn't notice the directory change. I'll fix that.
John Boyer: We'd like to have a resolution to submit XForms 1.0 TE to PER. It's the same document that's been available for six months now except for the change in status of the version attribute. So I take that as resolved.
Resolution 2007-07-18.2: We'd submit XForms 1.0 TE to PER.
John Boyer: The Transition Request is time-consuming.
John Boyer: Michael Kay will be
unable to be the keynote speaker due to a conflict.
John Boyer: Also time is short for
proposals. We have Mark Seaborne, Charlie, and a third one. We
could do eight 15-minute, six at 20, or six at 15 plus a 30 minute
keynote. Can you make it?
Steven Pemberton: No, Monday the 4th,
I can't make it. December 5th is like Christmas in Amsterdam.
John Boyer: What is the diff with
issue 106? Can we trash this one?
Steven Pemberton: I think we resolved
this already. Didn't I make the change?
Resolution 2007-07-18.3: http://htmlwg.mn.aptest.com/cgi-bin/xforms-issues/Appendices?id=151;user=guest;statetype=1;upostype=-1;changetype=-1;restype=-1 was a reply to 106 and Steven deleted it.
John Boyer: The terminology of verb
became less enchanting when I looked around for some normative
definition of verb. Eric posted a suggestion that we reconsider
manipulating the method attribute of submission instead, which we
had originally considered. The submission method says things like
post or get or urlencoded-post, and was a bit of an abstraction.
One of the problems that we had was that method also implies the
serialization; with post, you get xml serialization. We were
concerned at the time that if we used an arbitrary method, that it
wasn't clear what serialization would be implied. During the course
of this week, we realized that we've come to accept that an
attribute can have a default based on the value of another
attribute; so a serialization attribute on a submission defaulted
to application/xml if the method=post, and for
method=get defaulted to application/x-www-... and then
for some other string, it would be application/xml. If we had a
serialization attribute as well and get rid of the verb idea.
Leigh Klotz: There are two problems
that will come up; the keywords we use carve out of the space but
we don't have a name, and the second is that HTTP uses uppercase,
and ours use lowercase, so something has to know to convert.
Steven Pemberton: So what about saying
other methods are uppercase.
Leigh Klotz: We should have answers to
these questions though.
John Boyer: If there were a use case
other than HTTP then we would see we needed to solve the
problem.
Steven Pemberton: I think we're
probably OK. Maybe not. Our fixed values are defined only for a
http, file, and mailto. Any other new value of method we don't
define anyway.
John Boyer: There is a schema binding.
The HTTP binding we define. The file binding doesn't use them all,
such as post.
Leigh Klotz: A lot of user agents
treat file: POST as PUT for debugging ease.
John Boyer: Fair enough. Anyway, the
default is almost always application/xml. There are
get and urlencoded-post that produce the
application/x-www-form-urlencoded.
Steven Pemberton: They really are
namespaced, but just according to the schema on the action
url.
Leigh Klotz: What do we do with the
prefixed names then? Do we pass them through?
John Boyer: Let's talk about that in a
minute. If we do have a serialization attribute then we don't need
a serialize attribute. We could have serialization="" MIME
type.
Steven Pemberton: That sounds great.
Erik had said he wanted to use the word NONE, but that's not a MIME
type.
Leigh Klotz: On the whole I like it
but we need to decide what to do about the prefixed methods.
Leigh Klotz: Wait, wasn't verb an
XPath expression?
John Boyer: Yes, so we need a method
child element.
Leigh Klotz: And also a serialization
child element.
John Boyer: Yes. That sounds like all
of a piece then.
John Boyer: Then I'll proceed to
create a spec-ready version for us to look at and get Erik to look
at it.