John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Steven Pemberton, CWI/W3C
Uli Lissé, DreamLabs
Charlie Wiecha, IBM
Steven Pemberton: Who owns the test
suite?
Nick van: The group.
Leigh Klotz: Anyone with CVS
access.
Steven Pemberton: Someone reported
that some of the test suite files aren't valid XML.
Nick van: Also the 1.0 test suite. The
test file close the link elements.
Steven Pemberton: The tests are fine.
It's the pages that point to the tests.
Nick van: I can put it on my list, but
I can't work on it right now.
John Boyer: It sounds like it could go
on the list behind other things.
Nick van: It isn't that much work, I
guess.
Uli Lissé: I can have a look at
them.
Action 2009-04-29.1: Uli Lissé to convert 1.1 test suite drivers to XHTML.
Uli Lissé: Nick proposed
hosting the test suite on Google Code.
John Boyer: We discussed it a bit;
we'd need to look at the license.
Leigh Klotz: I think this is the
license:
http://www.w3.org/Consortium/Legal/2008/04-testsuite-license.html
Nick van: Our test suite isn't under
that license; we haven't made that change.
John Boyer: I'd assumed anything from
the W3C would be covered under it.
Nick van: I saw a message a few months
ago that they were asking for a reference; there were a couple of
possibilities, including the link Leigh sent.
John Boyer: So just this needs to go
on the header page? Every file?
Nick van:
http://www.w3.org/Consortium/Legal/2008/04-testsuite-copyright.html
Leigh Klotz: I see, it says we have to
decide which license; BSD or W3C Test Suite.
John Boyer: The 3-clause BSD license
would make it easy to put up on Google Code somewhere.
Leigh Klotz: Either I think would
allow direct copies; the question is alterations. Since we expect
the test suite to be altered for testing, the BSD license seems
appropriate, but we really want only the non-XForms parts to be
altered, for host language integration. So it's a shame really
there's nothing else.
John Boyer: The W3C Test Suite License
says ...
Leigh Klotz: I see; it's not us who
gets to choose, it's the licensee: "The choice of license is up to
the licensee for every single use of tests from a W3C Test
Suite."
Leigh Klotz: It says how to use both
for test suite authors: "For either license, include the following
statement in a Test Suite (HTML markup shown):"
John Boyer: What constitutes
modification? Modifying the XForms components? I know we've had the
situation where the XForms model is in the head in some and in the
body in others.
Leigh Klotz: That's the point I was
making.
John Boyer: There's a difference
between modifiying the test and the rest.
Leigh Klotz: Right, it says "Test
Suites may distinguish the test harness (or, framework for
navigation) and the actual tests. For these Test Suites, the
prohibition of change only applies to the tests." So we're
covered.
John Boyer: So we need this markup o
the front page of our test.
Leigh Klotz: It just says "include the
following statement in a Test Suite" so just one copy seems
fine.
John Boyer: So if we use a repository
other than W3C does W3C care?
Steven Pemberton: I don't know. What
is the benefit of doing it this way?
John Boyer: Easier access.
Nick van: There's no CVS reporting,
but Google Code hosting gives us better logs.
Steven Pemberton: Do we have any
problem with working on it there and keeping a copy at W3C?
Nick van: I don't mind.
John Boyer: So moving snapshots to W3C
for specifications. So the W3C CVS isn't giving the reports you
want? The files aren't just in a zip file.
Nick van: Normally ViewVC or some
other tool for queries; I thought it wasn't.
Leigh Klotz: They used to have
something.
John Boyer: jigedit
Nick van: It only shows the last
files. I wasn't able to generate a report of what was
changed.
Leigh Klotz: Do you have command line
access?
Nick van: It's Eclipse. I can't get
Eclipse to give me history.
Leigh Klotz: So can Eclipise get to
W3C CVS?
John Boyer: I think it's just an
Eclipse problem. The command-line one works.
Nick van: I want to get to the top
diretory and see all changes.
John Boyer: It does seem like a lot of
work to move it out and back.
Nick van: When I have time again I'll
look into command-line CVS.
John Boyer: For 1.1, the hope is that
it won't need much more history.
Nick van: I'm not completely sure if
we captured everything. Keith and I both did changes and I want to
make sure we didn't lose any.
Leigh Klotz: Is there anonymous
readonly access? That's all we need for history.
Nick van: That would be awesome; ant
can generate XML history.
Steven Pemberton: Let me look.
Steven Pemberton: does putting ,cvslog
at the end of the URL work for you?
John Boyer: Why does it need to be
anonymous access?
Nick van: For read with ant, without
setting up a certificate. ant cvslog.
Leigh Klotz: What's the path to the
CVS server?
Nick van: I'll send it to you.
Action 2009-04-29.2: Leigh Klotz to investigate CVS ant log for W3C test suite.
Steven Pemberton: I have to ask for it for you. You have to send me a public key.
Action 2009-04-29.3: Nick van den Bleeken to attach W3C test suite license markup to root page of HTTP 1.1 http://www.w3.org/Consortium/Legal/2008/04-testsuite-copyright.html
John Boyer: The difference between
the test suite harness and the tes suite is also important. We want
our own host language, for example.
Leigh Klotz: So the "harness" is
anything not in the XForms namespace.
John Boyer: Right.
Leigh Klotz: And for XForms for
HTML?
John Boyer: That will be the normative
markup.
John Boyer: We have a report that
highlights rows without two implementations, but non-required
features need only one.
Nick van: We have normative-Basic and
normative-Full. Do we need it to be false in both?
John Boyer: The Full-vs.-Basic
identifies feature sets. The conformance section of 1.1 has matured
and uses "should" and "may" and "recommended" and "optional" any we
don't need a Basic Conformance Profile any more.
John Boyer: Consider these:
11.9.p mailto multipart-post submission (have none, need one?)
Nick van: Which ones should I
change? Ones not for normative-full need only one
implementation?
John Boyer: I'm not sure if all of the
tests have been correctly marked that way. So if the feature's not
required, we only need to have one. I don't know if all the tests
have been marked.
Nick van: They have, but maybe it's
not accurate.
John Boyer: So the rule would be if
it's not required, highlight the row only for zero
implementations.
Nick van: That's easy.
John Boyer: 11.9.p would not
highlight, but 11.9.q would highlight. 11.9.4.b would also
highlight.
Action 2009-04-29.4: Nick van den Bleeken to update rules for highlighting colors of XForms 1.1 reports to correctly handle non-required tests; check 11.9.p, 11.9.q, and 11.9.4.b afterward.
Nick van: There are some section 8
orange.
John Boyer: I'd like to go back to the
top of the list.
John Boyer: It sounds like that test has been removed.
Action 2009-04-29.5: Nick van den Bleeken to remove test 3.3.1.c3 from test report.
John Boyer: I think the reported
implementation was Chiba.
Nick van: Firefox 3.
John Boyer: Right.
Nick van: Mark had some
comments.
John Boyer: The Ubiquity processor
fails the test even though we dispatch xforms-model-destruct event
and the message happens. The test has replace=all and
there is a message. The level setting on the message is modal. We
think a lot of web-based implementations will have difficulty with
this test; it's hard to make something modal when the processor is
going away.
Nick van: ...
John Boyer: Is there another way to
test this? A setvalue?
Leigh Klotz: Could you do another
submission?
John Boyer: SO It could run
externally?
Leigh Klotz: Test it externally.
John Boyer: Like write to a
file?
Leigh Klotz: Or post.
John Boyer: You can do a
submission.
Leigh Klotz: So the page could ask for
a UID and then when you replace the page will have the UID.
John Boyer: It could show the
UID.
Leigh Klotz: You could populate the
test input with a UID and the POST could set it up.
Nick van: Or you could generate a
UID.
John Boyer: The page that comes back
has to do the query after, with a button.
Nick van: Selenium can automate
that.
John Boyer: It can be done on
xforms-ready.
Nick van: Or in XHTML.
John Boyer: So replace all happens; it
asks for a page; that page does a model-destruct; the
model-destruct submits a new uid; the page gets a new UID.
Leigh Klotz: The page it asks for has
an instance request, containing a UID.
Leigh Klotz: Now this is cross-domain
POST test.
John Boyer: That's a headache for
testing.
Leigh Klotz: So host the script the
same place that you host the test.
John Boyer: I think we allow domain
submission to happen instead of hosting the test; I'd have to ask
Mark Birbeck.
Leigh Klotz: Then this won't cause any
problems for you, and you can run the script somewhere else if you
want.
Leigh Klotz: I'll write the test
for 4.2.4.a. Does this work in Firefox?
John Boyer: Yes. The current one
works.
Action 2009-04-29.6: Leigh Klotz to re-write 4.2.4.a test to use submission in xforms-model-destruct and unique id to verify submission.
John Boyer: I thought it was required. I see it's "should." The table's right. It's covered by Nick's action to change the table rules.
John Boyer: Ubiquity doesn't
support upload. Does Chiba suppor it?
Leigh Klotz: Yes.
John Boyer: But not this. Looks like
EMC reported it.
Leigh Klotz: What does it do?
John Boyer: Good question.
Nick van: Why does Firefox pass upload
but not this one.
Leigh Klotz: It gets an
xforms-value-changed event when you select the file. I tried it in
the Firefox plugin.
Leigh Klotz:
http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt08/8.1/8.1.6/8.1.6.b.xhtml
John Boyer: Would you ask him if this
works in Firefox?
Action 2009-04-29.7: Leigh Klotz to ask Keith Wells if it's ok to mark http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt08/8.1/8.1.6/8.1.6.b.xhtml as passed.
Leigh Klotz: Done. http://lists.w3.org/Archives/Public/public-forms/2009Apr/0019.html