Alain Couthures, AgenceXML
Erik Bruchez, Orbeon
John Boyer, IBM
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Philip Fennell, MarkLogic
Steven Pemberton, CWI/W3C (chair)
Dan McCreary, Dan McCreary Associates
Uli Lissé, DreamLabs
Leigh Klotz: John, did my answer
satisfy you?
John Boyer: Yes.
Steven Pemberton: Which
proposal.
Leigh Klotz: So one question is what
about Cancel a Submission? It may be sensitive to cancel a purchase
that's already gone forward, but if we have these IDs you can just
ignore them yourself.
John Boyer: But you can cancel with a
browser button now, and nobody thinks it gives back your
money.
John Boyer: It seems useful to be able
to cancel a submission that's on the way out still.
Steven Pemberton: OK, there are
three proposal. In Proposal 1 where does the number come from? Does
the user make it?
Leigh Klotz: That's propoasl
1.1.
John Boyer: So the number connects the
xforms-submit to the xforms-submit-done or
xform-submit-error.
Steven Pemberton: And you keep
it?
John Boyer: Yes, you create a new
piece of data and put the transaction ID into an attribute with an
xforms-submit-done handler that uses an XPath predicate to put the
answer in the predicate. I think that's the basic idea. There's a
bit of an efficiency issue; the typical response we have a way of
targeting where it goes. It almost like it has to go there and then
you copy it somewhere else.
Leigh Klotz: You could have an XPath
function that in targetref that gives you the target number.
John Boyer: So that would just be the
event function.
Steven Pemberton: That sounds
good.
John Boyer: We'd have to call it out
in the spec that the evaluation of the targetref would have the
event in the context. The problem is that the XPath event function
is defined to work only when you process an event, whereas
targetref is evaluated before the event happens. So it would have
to be some other function. The targetref is evaluated, the data is
replaced, and then the xpath-submit-done happens.
Leigh Klotz: Or you could have another
event.
John Boyer:
xforms-submit-nearly-done
Leigh Klotz:
xforms-submit-replace-instance
John Boyer: A couple more events to do
the same work.
Leigh Klotz: The advantage is you'd
have the same function in both places.
John Boyer: That's in keeping with the
roadmap for uploading, in-progress, etc.
Leigh Klotz: We should be consistent
with XHR's lifecycle.
Steven Pemberton: We should perhaps
code these out and see how they look.
Leigh Klotz: I'm not a big fan on
1.1.
John Boyer: There may be headaches
with 1.0. We copy data from A to B.
Leigh Klotz: These are all additive
proposals.
Steven Pemberton: We have a new
proposal as well.
Leigh Klotz: That'a additive as well,
adding a new event.
Steven Pemberton: So how about
1.2?
Leigh Klotz: Oh, right, it's built on
1.1 as well.
Steven Pemberton: What are the next
steps? Flesh out proposal 1 and do example code.
John Boyer: I'm not pushing for
this.
Steven Pemberton: What are the use
cases?
Leigh Klotz: Multiple submissions for
the same submission element. Multiple outstanding
submissions.
Steven Pemberton: Who else does
this?
Leigh Klotz: YUI. The predecessor to
Android did this.
Steven Pemberton: Ok, you can fill in
more.
ACTION-1794 Leigh Klotz to fill in http://www.w3.org/MarkUp/Forms/wiki/Submission_Request-Response_IDs Proposal 1, add the new xforms-submit-replace-instance event, and write sample code.
Leigh Klotz: There are some
specific questions and some general questions.
Steven Pemberton: I think he's using
Kurt Cagle's example http://xformstest.org/2010/07/kurt_cagle/index.xml
Steven Pemberton: He says "lose
control of the tab."
Leigh Klotz: You can click below the
tab and it loses focus. Also, setvalue. But John's switch/@ref
would help with setvalue.
John Boyer: This isn't repeat: this is
switch. Our switch is slightly underpowered because we say only one
case is available and the others are not. Unfortunately the label
for the case lives inside the case. So there's been confusion. To
me that's just an appearance on switch, tabbed-appearance. The
label inside the case is metadata. The switch can display it
however it wants. Availability and relevance are about the form
controls inside a case, not the metadata.
Leigh Klotz: How would you control
whether a case is available or not?
John Boyer: The selected case is the
only one available. People have the reaction to making it a trigger
so they can run a toggle. That's where we run into trouble.
Leigh Klotz: You have a switch/@ref
proposal. Can you use relevance to control which tabs are
available?
John Boyer: You could use setvalue
instead of toggle, but you still need an action or data to change
which case you are on. That's a separate issue from describing
which data are available. XForms switch can still display only the
cases in the UI. We don't have that.
Leigh Klotz: You had switch/select1
earlier in your proposal.
John Boyer: The switch from select1
would populate the data.
Leigh Klotz: I meant the proposal of
combining the two, not using them together. But we lost something
when we went to switch/@ref.
John Boyer: Something like a tabbed
interface would be outside the switch element or a group of
triggers on top. What we lost was that the switch would be the
rounded square, and the four tabs above it would be separate
controls.
Leigh Klotz: That approach works well
if we have a component system, but without it we're doing a lot of
repetitive typing.
John Boyer: Absolutely. We'd like to
make repeat smarter and provide tabbed interfaces, but haven't
gotten around to it.
Steven Pemberton: Right, we haven't
crystallized the changes we want to make to be able to fully
describe it. So Michael's problem here is sort of at a presentation
level. He can do the functional part, but he can't get the
presentation he needs. The question is whether any of the proposals
we have at hand would solve his problem.
Leigh Klotz: What presentation?
Steven Pemberton: The examples work
but the color of the tabs change in
http://www.w3.org/MarkUp/Forms/2010/xforms11-for-html-authors/tabs.xml
John Boyer: As you press the title the
file description it turns grey.
Dan McCreary: Which one?
Steven Pemberton:
http://tei2010.blackmesatech.com/exx/helloworld/cagle_tabs_mod.xhtml
Dan McCreary: When you click in the
content area, the tab label goes gray.
Steven Pemberton: It uses CSS focus.
CSS wasn't designed for this. So in a sense that's too bad, but the
essential question is, can he do it in some other way, or are we
planning something?
Dan McCreary: Don't the existing
examples in the wikibook work?
Dan McCreary:
http://xforms-examples.googlecode.com/svn/trunk/02-controls/10a-Horiz-Tab-Menu/XForms-Horizontal-Tab-Menu.xhtml
John Boyer: The markup example shows
that he is able to do a tabbed interface for switch without
switch/@ref. If this switch did use data referencing all that would
be different is that when you toggle between cases, the active case
would be stored. But this construct would work the same.
Dan McCreary: It uses the CSS target
attribute.
John Boyer: Dan, does that do what
Michael wants?
Dan McCreary: I use the target
pseudo-element.
Steven Pemberton: Which one?
Dan McCreary: The example doesn't
match the text. http://en.wikibooks.org/wiki/XForms/Horizontal_File_Tab_Menu
John Boyer: He wants the selected tab
to be highlighted.
Dan McCreary: That's different; mine
has a hard-coded tab color.
John Boyer: Yes, that's not the effect
he wants.
Steven Pemberton: I see how the target
property could work, but it messes up with the history. Every time
you click, it goes back to the previous tab you are on.
Dan McCreary: That is perhaps an
unwanted side effect.
Steven Pemberton: It's an interesting
effect.
Leigh Klotz: The web-apis Javascript
history functions may make this easier.
Steven Pemberton: Sounds like a big
hammer.
Dan McCreary: We should have a best
practice for tabs.
Steven Pemberton: The problem seems to
be the unrolling. Normal toggles will do the rigth thing. Isn't the
problem the technique.
John Boyer: I don't think so. Even
when Dan's technique, you can't get the technique to work where all
non-selected cases are the other color.
Nick van: I use AVT for this. It's
easy. You can change the class of a div based on instance
data.
Dan McCreary: Do you have a
sample?
Nick van: It's quite easy; just an
XPath if() function: {if(current-selected-tab=current-value, then
selected else non-selected).
Dan McCreary: I'd like to write up
examples.
Steven Pemberton: It's not official
XForms yet.
John Boyer: Does this example use the
@ref version of switch?
Nick van: No, it uses group and
repeat. There is an element for every repeat item, same as Kurt
Cagle's first example. Then using AVT on the div of the label you
can inject the correct class on the div and style it in CSS.
Steven Pemberton: So what Michael's is
asking for is the ability to style the last-pressed toggle.
Leigh Klotz: If we had switch/@ref
then you wouldn't need the...
John Boyer: You wouldn't need the
setvalue and toggle would write the data. You can then style the
trigger based on which case you have.
Steven Pemberton: Yes.
John Boyer: People want to use data in
presentation calculations.
Steven Pemberton: Yes, you want
negative numbers to be red.
Steven Pemberton: Good discussion; no
particular solution, but I think we can answer Michael.
ACTION-1795 Steven Pemberton to answer http://lists.w3.org/Archives/Public/public-forms/2010Sep/0055.html