Re: [office-metadata] Groups - Metadata_Model_Proposal_23May2007(07-05-23-ODF-Metadaten.odt) uploaded

[Svante Schubert <Svante.Schubert@Sun.COM>]

John,

As Bruce already gave a perfect explanation why text:span should not be used, I want to add that our alternative to text:span is the text:meta element.
Find an example, a comment on your found issue and finally a further issue below.

John Madden wrote:

7F21D752-2602-4B3B-9BBB-D5E23DF4BE42@mac.com" type="cite">

Bruce,

Good you told me that, and I'm sorry I'm not entirely up to speed on this.

So would text:bookmark be the ODF equivalent of an xhtml:span?

Could you guys just send me a quickie example of what you figure the best practice is for creating a hook inside of text to hang an xml:id onto?

I have a little problem with ODF bookmarks, because using OpenOffice as an experimental platform, I get this:

<text:p text:style-name="Standard">

this is a length of otherwise un-marked-up text, but I want a 

<text:bookmark-start text:name="hookMark"/>hook<text:bookmark-end text:name="hookMark"/> 

to hang an xml:id on the word “hook”.

</text:p>

The above is valid ODF and used in OOo, in ODF 1.2 you could add an xml:id.

<text:p text:style-name="Standard">

this is a length of otherwise un-marked-up text, but I want a 

<text:bookmark-start xml:id="foo" text:name="hookMark"/>hook<text:bookmark-end text:name="hookMark"/> 

to hang an xml:id on the word “hook”.

</text:p>

But unless you do not really need the feature of a bookmark or your text breaks XML boundaries like

<text:p text:style-name="Standard">

this is a length of otherwise un-marked-up text.

<text:bookmark-start xml:id="foo" text:name="hookMark"/>example

of the word “hook”.

</text:p>
<text:p text:style-name="Standard">

The word "hook" is very important!<text:bookmark-end text:name="hookMark"/> 
        Identified by the xml:id on the selected text.

</text:p>

it should be the common way to use text:meta for such an example in ODF 1.2:

<text:p text:style-name="Standard">

this is a length of otherwise un-marked-up text, but I want a 

<text:meta xml:id="foo">hook</text:meta> 

to hang an xml:id on the word “hook”.

</text:p>

7F21D752-2602-4B3B-9BBB-D5E23DF4BE42@mac.com" type="cite">

This highlights that bookmarks in ODF don't work like <xhtml:span>yadda-yadda</xhtml:span> elements, but instead there is a <text:bookmark-start/> and <text:bookmark-end/>

Do if I want to assert some metadata about the word "hook" at this place in the document I have a little issue, namely:

This seems a little problematic from the point of view of using them as a way of referencing into text, because now my xml:id would normally be taken by your plain-vanilla xml parsers to refer to just the text:bookmark-start element, rather than to the piece of the document that includes the word "hook".

7F21D752-2602-4B3B-9BBB-D5E23DF4BE42@mac.com" type="cite">

Solutions?

7F21D752-2602-4B3B-9BBB-D5E23DF4BE42@mac.com" type="cite">

John

Yes, you are right the selection is something different than the content of the stand-alone text:bookmark-start element.

There are two spots we have to reconsider wording 1.2.2 mentioning "...the literal content of the OpenDocument element" and 1.1.1 "...the content literal of the OpenDocument element the RDF object."

"1.2.2 Declaration of In Content Metadata

The following attributes are used for in content metadata:

• m:about : The IRI of the RDF subject

• m:property : The IRI of the RDF predicate

• m:data-type: Data type of m:data-value

• m:data-value: The RDF object, if it appears. Otherwise, the literal content of the OpenDocument element is the RDF object."
  

and beyond chapter 1.1.1

"For every IRI representing an OpenDocument element an RDF statement can be made about the element content. In this RDF statement the element IRI is the RDF subject, the property of odf:content is the RDF predicate and the content literal of the OpenDocument element the RDF object."

We could make it more explicit by adding the case where the content between the text:bookmark-start and text:bookmark-end is meant.


A further issue is that currently even an empty text:bookmark element is allowed to use the in content metadata. It makes no sense to allow a m:data-value when there is no content.

Svante

7F21D752-2602-4B3B-9BBB-D5E23DF4BE42@mac.com" type="cite">

On May 23, 2007, at 5:21 PM, Bruce D'Arcus wrote:

On May 23, 2007, at 6:07 PM, John Madden wrote:

Wanted to make sure that xml:id is allowed on <text:span>.

Is that correct?

I don't know the answer, but keep in mind that ODF's definition of span is somewhat unique. It's basically a structure to apply presentation to, rather than having any fundamental semantics.

So as Michael has explained it, these two are formally equivalent from an ODF perspective:

<text:span>some text</text:span>

<text:span>some </text:span><text:span>text</text:span>

... and applications are fine to treat them as such (equivalent).

The consequence, it seems to me, is that unless the TC cares to redefine their meaning, spans are not a reliable target for metadata.

Bruce