XHTML2+XForms attribute clashes [was: RNG Schemas for XForms]

(As noted in: http://www.w3.org/2009/03/04-forms-minutes.html#item01)

During working on integration of XForms into XHTML2 in RelaxNG, an
issue of attribute name/semantic clashes came up. The below is a
summary of the problem (which, obviously, applies equally across
schema languages).

XHTML2 adds the XHTML2 Common attribute collection to the XForms
Common Attribute Group[1]. The following attribute collisions occur.
(Note - I cannot guarantee that the below is 100% correct, as we have
not yet ascertained that the current XHTML2 XForms module draft (based
on Leighs RNC work) is in full compliance with the latest greatest
XForms 1.1 draft. Any corrections warmly welcome).

(notation: xforms location[datatype] --> xhtml2 attribute collection[datatype])

xf:instance/@resource[anyURI] --> rdfa:@resource[URIorSafeCURIE]
xf:submission/@resource[anyURI] --> rdfa:@resource[URIorSafeCURIE]
xf:submission/@encoding[String] --> embedding:@encoding[Encodings]
xf:submission/@target[xforms:XPathExpression] --> hypertext:@target[HrefTarget]
xf:dispatch/@target[IDREF] --> hypertext:@target[HrefTarget]
xf:load/@resource[anyURI] --> rdfa:@resource[URIorSafeCURIE]

A brief off-list discussion was held including Leigh, Shane and
Steven. Heres a quick recap:

First solution suggestion: for each collision, make the host not
inject the particular colliding xhtml attribute into the particular
xforms element. Also consider if there are datatype deviations that
are so small that they can be ignored.

But, as a potential objection to the above, consider the result from
an end user point of view:
- In previous collision spots, these attributes would be available
with semantics as defined by XForms;
- all other elements in XForms would also carry these attributes, but
with semantics as defined by XHTML2.
- and because we do not namespace-qualify attributes, there is no
shallow (user friendly) way to say which is which (think: in a popup
menu of a context sensitive schema driven XML editor).

Other solution suggestions:
- making a generic modification to the XHTML family conformance
requirement that a module *must* include the entire Common attributes,
and relax it into to a *should*. Allow the injection of a subset of
Common (eg Core, I18n, ?) as an alternative, and make the other
sub-collections in xhtml.Common injection-optional.

- make a specific deviation from the above rule in the case of XForms
(and include only a non-colliding subset of xhtml.Common for the
entire xforms element set). Solves the problem in the particular case
of XHTML2+XForms, but doesnt solve the problem architecturally.

- examine if we would be able to enforce the current XHTMLMOD
recommendation that xhtml attributes in a foreign module should be
specified using a prefixed QName (eg <xf:submission xhtml:target="foo"
target="bar">).

hth, /markus

[1] http://www.w3.org/MarkUp/2009/ED-xhtml2-20090205/mod-xforms.html#XForms_Other_Attrs

On Tue, Jan 20, 2009 at 6:22 PM, Klotz, Leigh <Leigh.Klotz@xerox.com> wrote:
> Markus,
> Thank you for contacting us.  I did not use the XHTML M12N because I was
> unaware of it.
> The pub date on it is 2 weeks before my action was assigned, so it had
> not come to my attention.
>
> We discussed the issue at the F2F in Amsterdam in June 2008, and
> examined the ATOM RFC for its use of RNC, but none of us was aware (to
> my recollection) of this document.
>
> We have no particular attachment to the document I sent; re-use is not
> necessarily a goal.  (Personally, I will be working to make sure that
> whatever gets produced works well with James Clark's NXML package, but
> that's a separate issue.)
>
> So, we would be pleased if you would take on the first draft of the
> XForms 1.1 work.  Our XForms 1.2 M12N work is still in progress, and
> coordinating with XHTML2 WG on that is beyond the scope of my action;
> however, I'm sure the Forms WG would be interested in further
> discussions on using xhtml-m12n-relaxng as a base.
>
> The M12N buckets we have been working on are here:
> http://www.w3.org/MarkUp/Forms/wiki/XForms_1.2_Modularization
>
> There are some pre-draft documents for the modules as well, but they are
> in flux.
>
> Leigh.
>
>
> -----Original Message-----
> From: Markus Gylling [mailto:markus.gylling@gmail.com]
> Sent: Monday, January 19, 2009 4:44 AM
> To: Klotz, Leigh
> Cc: public-forms@w3.org; shane@aptest.com
> Subject: Re: RNG Schemas for XForms
>
> Hi Leigh,
> thanks for providing this.
>
>>This message is both a response to the XHTML2 WG request for XForms RNG
>
>>schemas, and request in support of my own action to seek expert help.
>
> I have just started working on updating the draft XHTML2 RNG schema,
> including its XForms support. While doing this I will review your work,
> obviously targeting reuse. As for your request for a review of design
> patterns etc, I am hoping to be able to get back to you with a set of
> suggestions quite shortly.
>
> Leigh, a question: while writing the modularized schema that you
> forwarded, did you read and/or make use of [1]? In that case, were there
> (aside from group patterns that you've already mentioned) aspects of
> RelaxNG schema design that you found insufficiently discussed? We are
> planning to update this doc as well in the not too far future, and any
> input on its nature and coverage is warmly welcome.
>
> [1] http://www.w3.org/MarkUp/2008/ED-xhtml-m12n-relaxng-20080604/
>
> Best regards,
> Markus Gylling (new XHTML2 WG member)
>
>
>>>>>>>>>>
> Shane,
>
> Forms WG Action 2008-06-18.1
> <http://lists.w3.org/Archives/Public/public-forms/2008Jun/att-0050/2008-
> 06-18.html#ACTION1> calls for me to produce a modularized XForms RNG
> schema.
>
> I produced a version of it, based on original work by Micah Dubinko, and
> posted it in this thread:
> <http://lists.w3.org/Archives/Public/public-forms/2008Jun/0054.html>
>
> I found I was unable to complete the task because I lacked an
> understanding of design patterns for RNG groups.  Accordingly, Action
> 2008-07-23.3
> <http://lists.w3.org/Archives/Public/public-forms/2008Jul/att-0040/2008-
> 07-23.html#ACTION3> calls for me to seek expert help.
>
> Additionally, Steven Pemberton today reported that the XHTML2 WG
> requested RNG schemas for XForms.
>
> This message is both a response tp the XHTML2 WG request for XForms RNG
> schemas, and request in support of my own action to seek expert help.
>
> Attached is a zip file containing my best effort for the XForms 1.2
> modularization.
>
> For historical reasons, this zip file is called xforms-1.2.zip.
>
> In the zip file is also xforms11.rnc file from Micah Dubinko, with my
> modifications for re-hosting the grammar as part of James Clarke's RNC
> package of schemas for XHTML distributed as
> <http://www.thaiopensource.com/nxml-mode/>
>
> The original XForms RNC files are from Micah's XFV project, at
> https://sourceforge.net/projects/xfv/
>
> When I sent some corrections to Micah, he made me an administrator of
> the project, so I will be able to update the project with the results of
> this work.
>
> As reported in
> <http://lists.w3.org/Archives/Public/public-forms/2008Jun/0055.html>,
> Micah Dubinko has graciously offered to let us use the XForms 1.0 RNC
> schema he produced, and is prepared to relicense it if we need to have
> it under the W3C Software License instead of the Apache license.
>
> This schema for XForms 1.1 is current as of its original publication
> date, but XForms 1.1 has changed since then.
>
> Please let me know if you'd like to proceed on this issue.
>
> Thank you,
> Leigh.
>

Received on Thursday, 5 March 2009 10:52:37 UTC