Re: idValue issues

2012/10/14 Shaun McCance <shaunm@gnome.org>

> On Sun, 2012-10-14 at 12:47 +0200, Felix Sasaki wrote:
> >
> >
> > 2012/10/13 Shaun McCance <shaunm@gnome.org>
> >         So although all the *Pointer attributes are defined to be
> >         relative
> >         location paths, I think we just end up getting a string value
> >         from
> >         the pointed-to nodes for everything except targetPointer.
> >
> >
> > No -
> >
> http://www.w3.org/International/multilingualweb/lt/drafts/its20/its20.html#EX-locNotePointer-attribute-1
> > here you get the complete notes element
> > <notes>Indicates that the resource file {0} could not be
> > loaded.</notes>
> > since the XPath expression of the pointer attribute is ../notes
> > you would get several "notes" elements, if available.
>
> Right, but what do implementations actually *do* with that information?
> I convert each pointed-to node to a string and serialize that into a PO
> file. Does anybody else do something that doesn't involve stringifying?
>

If the node pointes to a loc Note that has some inline markup, e.g.

<p>Read <a href="#explanation">this explanation </a> before starting the
translation</p>

I think it is crucial not to do stringifying.


>
> Plus, these attributes tend to be defined with language like "contains
> a relative selector pointing to a node that holds..." A node, singular.
> What am I supposed to do if langPointer returns two nodes?
>


What do you do if the XPath expression in the "selector" attribute selects
nothing? With XPath or any selector mechanisms there are always cases that
don't make sense. And these cases depend both on the XPath expressions and
the content. But that's a different issue IMO than the stringification of
the relative location paths. You won't win be forcing the stringification.

Felix


>
> --
> Shaun
>
> >
> >         In fact,
> >         I could imagine wanting to do something like this with params:
> >
> >         <its:rules>
> >         <!--
> >         For in-house translators, instead use:
> >         $baseurl = http://internal.example.com/docs/
> >         -->
> >         <its:param name="baseurl">http://example.com/docs/</its:param>
> >         <its:locNoteRule selector="//*[@noteref]"
> >           locNoteRefPointer="concat($baseurl, @noteref)"/>
> >         </its:rules>
> >         <p noteref="notes.html#foo">Foo</p>
> >
> >         That example doesn't conform, but it seems legitimate.
> >
> >         --
> >         Shaun
> >
> >         On Mon, 2012-10-08 at 08:35 -0600, Yves Savourel wrote:
> >         > Hi Jirka, all,
> >         >
> >         > I believe the idValue attribute is not the same as the
> >         classic
> >         > pointers we have in other places.
> >         >
> >         > The definition says: "It contains an XPath expression which
> >         constructs
> >         > a string corresponding to the identifier of the node to
> >         which this
> >         > rule applies[ should be located]."
> >         >
> >         > (Note there is a copy/paste error: the words "should be
> >         located" at
> >         > the end should be deleted. I'll fix that).
> >         >
> >         > So the value returned by the XPath expression is not a node
> >         where to
> >         > find the Id value, but the ID value itself. This allows to
> >         construct
> >         > IDs like shown in example 63
> >         >
> >         (
> http://www.w3.org/International/multilingualweb/lt/drafts/its20/examples/xml/EX-idvalue-element-2.xml
> )
> >         >
> >         > -yves
> >         >
> >         > -----Original Message-----
> >         > From: Jirka Kosek [mailto:jirka@kosek.cz]
> >         > Sent: Monday, October 08, 2012 8:16 AM
> >         > To: public-multilingualweb-lt@w3.org
> >         > Subject: idValue issues
> >         >
> >         > Hi,
> >         >
> >         > I have just noticed two inconsistencies in definition of
> >         idValueRule
> >         > (http://www.w3.org/TR/its20/#idvalue):
> >         >
> >         > - idValue attribute is defined as containing XPath
> >         expression. It
> >         > should contain "relative selector" in order to support new
> >         > queryLanguage
> >         >
> >         > - also should be this attribute named idValuePointer - on
> >         all other
> >         > rule elements attributes pointing to elements containg
> >         actual value
> >         > are suffixed with "Pointer"
> >         >
> >         > Thoughts?
> >         >
> >         >                               Jirka
> >
> >
> >
> >
> >
> >
> >
> >
> > --
> > Felix Sasaki
> > DFKI / W3C Fellow
> >
>
>
>
>


-- 
Felix Sasaki
DFKI / W3C Fellow