W3C home > Mailing lists > Public > public-i18n-its@w3.org > April to June 2006

The termRef, locInfoRef attributes

From: Yves Savourel <yves@opentag.com>
Date: Wed, 3 May 2006 08:03:15 +0200
To: <public-i18n-its@w3.org>
Message-ID: <003001c66e77$45356a90$640fa8c0@Breizh>

Hi all,

I had an action item to look more into the reference mechanism for termRef and locInfoRef. Both attributes are used to provide a
reference location where some information pertaining to the selected node can be found.

--- termRef is defined as "external information about the term". [BTW: I'm not sure about "external" as the information could be in
the same document as shown in example 23.]

--- locInfoRef holds a reference to a location where the localization information note is.

#1-- So, first I see a slight inconsistency with naming: "[term]Ref" references data about the term (not the term), while
"[locInfo]Ref" references the "locInfo" itself.

In theory "termRef" should be more like "termInfoRef" or "termDefRef". I wonder if we need to correct this or not. Keeping "termRef"
could be an issue if later we decide to provide a way to define the term using ITS (like a termInfo attribute or element).

#2-- URI. I've looked at various examples of referencing/linking mechanism with regard to separating the ID and the source document
where the ID is located, and it seems several assume the use of xml:base to specify any base URI. It's not quite the same as
specifying the source file, since xml:base could be only a partial path and the "ID" part would be a relative location for that

So, I suppose if you have locInfoRef="#IDabc", it points within the document where that attribute is only if there is no xml:base

Looking at xml:base, it seems to work a bit like xml:lang (valid for the element where it is and its children, except when overriden
by another xml:base. This makes things a bit complex for implementation.

Note: we do have to handle xml:base with our <its:rules xlink:href="abc"/> since XLink processing relies on xml:base

Relative path are generally a good thing for portable document. I would tend to have such mechanism, but I'm worried about the cost
of implementing xml:base.

#3-- The xml:base + href-type attributes serve a purpose different than splitting ID and the container of the data for the ID. I
think it's OK for locInfoRef, but for termRef it's a bit different. For example in TBX-Link 'id' holds an ID (e.g. "abc" not "#abc")
and 'termbase' holds a URL not a base path. So one could reconstruct this using xml:base and termRef, but it's different from just
associating an xml:base-like and href-like attributes.

I could not find many other formats that had term identification features (e.g. <term>) working with some kind of referencing to the
definitions, so far:
- TEI (http://www.tei-c.org/P4X/TE.html) but this is for coding a ternbase rather than marking up a document.
- DITA has a <term> with a keyref attribute (of type NMTOKEN), but keyref is said to not be implemented for now.

So to summarize:

- we may want to change "termRef" to something like "termInfoRef" or "termDefRef" (?)

- reference attribute + xml:base provide a base+relative path mechanism. Do we want to use this? Do we need to explicitly mention

- referencing with regard to terms is done differently in different formats. Our current termRef allows for URI association, but how
should we treat ID (this is the old question from Chritisan: what if the data to 'map' have different type?)

Any thought anyone?
Received on Wednesday, 3 May 2006 06:03:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:43:07 UTC