RE: [office] text:span elements for RDFa attributes

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

Svante,

Surely you mean for your examples to be

<text:span text:style-name="green">green</text:span>
<text:span text:style-name="red">red</text:span>
<text:span text:style-name="green">green</text:span>

<text:span text:style-name="green">green
<text:span text:style-name="red">red
</text:span>green</text:span> 

In any case, there is nothing to prevent use of a <text:span> element in
exactly the same way they are used in HTML and XHTML as a way to provide
embracing elements.  The RDFa specification is loaded with examples of the
creative use of <xhtml:span> as a way to provide for creative in-line
metadata, and the rules for finding and extracting it are also pretty clean.


It seems to me that <text:span> is no more volatile than in RDFa (your
inspiration, after all).

Likewise, since hierarchy must be observed, introducing <text:meta> as a
special <text:span> variant doesn't resolve the volatility issue.  I do
grant that it provides a nice non-backward-compatible way of knowing where
to look for RDFa attributes in text content.

I am bothered by the following text in OpenDocument-v1.2-draft7-13.odt
description for <text:meta> though:

      "The <text:meta> element represents a location in a document 
      where content can be generated from metadata and inserted in
      a document instance."

That is not exactly what we want from a span-type element.  On the other
hand, I don't see anything that suggests how the generation of content is
supposed to occur.  Perhaps this description is a mistake? 

Now what we have is two elements that each can occur everywhere that the
other can and the only difference is the attributes they can carry.  It is
an interesting nuance that one can impact style and the other only presents
metadata, but I wonder if that is worth having a non-backward-compatible
element which will presumably result in foreign-element behavior (as opposed
to the more-tolerable foreign-attribute behavior).  

I also wonder how much of RDFa applies to <text:meta> by implication.  That
is rather interesting and I confess that I overlooked the availability of
this element in my previous studies.  (It also reduces my concern about
absence of RDF support in the <office:document> case.)

Thanks, I completely missed this.

 - Dennis

PS: I don't think the difficulty of testing documents that are equivalent
but have different XML is unique to <text:span> and the <text:meta> can be
used in non-unique ways too, especially thanks to the wonderful variability
in RDF, the use of CURIEs, etc.  In fact, simple differences in use of
namespace prefixes will make literal comparison of generated XML documents
problematic.  

-----Original Message-----
From: Svante.Schubert@Sun.COM [mailto:Svante.Schubert@Sun.COM] 
http://lists.oasis-open.org/archives/office/200901/msg00113.html
Sent: Thursday, January 15, 2009 12:40
To: dennis.hamilton@acm.org
Cc: 'office'
Subject: Re: [office] Empty text:span elements

Dennis,

please find my answer to RDF usage on text:span below.

Dennis E. Hamilton wrote:
http://lists.oasis-open.org/archives/office/200901/msg00112.html
> Thomas,
[ ... ]
> PS: What thrills me about your note is that I am so unfamiliar with ODF
> details that I didn't know there was a text:span element at all.  Seeing
> that there is, I am now completely baffled why the RDF metadata proposal
> does not allow it for decoration of text with in-line RDF annotations in
> precisely the way that <span> is used in the RDFa extensions to XHTML.
> <text:span> is a wonderful feature for that, even though it is not as
> generously usable as in the HTML cases.  I had supposed that this simple,
> direct method for in-line annotation with metadata (a.k.a semantic markup)
> was not supported because there was no span element.  Fooled myself!
>   
A text:span is a quite volatile element in ODF, the following is equivalent:

<text:span text:style-name="green">green</text:style-name>
<text:span text:style-name="red">red</text:style-name>
<text:span text:style-name="green">green</text:style-name>

<text:span text:style-name="green">green
<text:span text:style-name="red">red
</text:style-name>green</text:style-name>

With splitting&joining spans, the target of the RDF would be changing.
In case you want to add semantic to a text:span, we added a new element
called text:meta.

The text:span behavior is as well a problem for testing, when comparing
ODF documents [1].

Regards,
Svante

[ ... ]