W3C home > Mailing lists > Public > public-multilingualweb-lt@w3.org > July 2012

Re: [ACTION-153] related to [ISSUE-15] targetPointer

From: Shaun McCance <shaunm@gnome.org>
Date: Mon, 09 Jul 2012 14:33:05 -0400
To: public-multilingualweb-lt@w3.org
Message-ID: <1341858785.2196.85.camel@recto>
On Mon, 2012-07-09 at 17:20 +0200, Yves Savourel wrote:
> Hi all,
> 
> I have the action item to summarize the status of the targetPointer
> requirements
> This related to ISSUE-15.
> 
> The current requirement
> (http://www.w3.org/International/multilingualweb/lt/wiki/Requirements#targetPointer) outlines several use cases.
[snip]
> Shaun was illustrating his thoughts with:
> 
>  selector="//license[@xml:lang='en']" repeatable="yes"

For the benefit of anybody else reading, I want to clarify that I'm
approaching this purely from the angle of constructing multilingual
documents. I don't personally have any need for identifying targets
in existing multilingual files.

> To describe a case like this:
> 
> <application>
>  <license xml:lang="en">
>   <p>This is the license.</p>
>   <p>It has multiple paragraphs.</p>
>  </license>
>  <license xml:lang="de">
>   <p>Dies ist die Lizenz.</p>
>   <p>Es verfügt über mehrere Absätze.</p>
>  </license>
> </application>
> 
> Such notation however:
> 
>  a) It would not work for cases where the target(s) are in non-sibling
> nodes.
> 
>  b) It would work only if the document uses xml:lang to distinguish
> the different language blocks (as it is the only implicit difference
> one can make in the example).

Both true. I have existing functionality that shares these limitations
and has its own extra limitation: in the case above, the <p> elements
would be repeated, unless <p> was marked as withinText=yes.

> <application>
>  <src>
>   <license>
>    <p>This is the license.</p>
>    <p>It has multiple paragraphs.</p>
>   </license>
>  </src>
>  <targets>
>   <target langCode='1'>
>    <license>
>     <p>Dies ist die Lizenz.</p>
>     <p>Es verfügt über mehrere Absätze.</p>
>    </license>
>   </target>
>  </targets>
> </application>
> 
> Or, more simply, for bilingual cases where the content is a single
> element:
> 
> <msgs>
>  <m id='1'>
>   <src>Text of the message</src>
>   <trg>Texte du message</trg>
>  </m>
> </msgs>
> 
> So, I think an XPath relative expression would be more flexible
> because it would allow both use cases:
> 
> <its:targetPointerRule "//license[@xml:lang=$SourceLang]"
> targetPointer="../license[@xml:lang=$Target]"/>
> 
> <its:targetPointerRule "//src"
> targetPointer="../targets/target[@langCode=$TargetCode]"/>
> 
> However, this method has also a possible drawback: like with Shaun's
> example, with more complex expressions creating the target node if it
> does not exist, may be complicated. 

Here's a concrete example of an expression I can't do anything with:

targetpointer="../*[local-name(.)!='src']/*[@langCode]"

Even with the example you gave, which I could parse, I'd have to make
some assumptions. Does <targets> come before or after <src>? Does each
<target> get its own <targets>, or are they all under one <targets>?
And if I'm not using xml:lang, what am I using, and how do I construct
the values?

> I wonder if then the data category scope could be limited to work, at
> least, with existing nodes?

I think we're looking at two very different sides of the same coin.
Unfortunately, I can't think of anything that solves both problems
well. The existing targetPointer proposal looks very sane for what
you're trying to do.


-- 
Shaun McCance
Community Help Expert   |   Open Help Conference
http://syllogist.net/   |   http://openhelpconference.com/
Received on Monday, 9 July 2012 18:33:29 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:31:47 UTC