[xliff] XLIFF: datatype-based approach to more complete lists ofpropose d values

["Lieske, Christian" <christian.lieske@sap.com>]

Dear all,

To me, the work that Yves, John R, Enda, Jaime have started wrt. to a common
way to represent Windows resources in XLIFF (see
http://lists.oasis-open.org/archives/xliff/200206/msg00026.html) is
extremely interesting. I already indicated this in
http://lists.oasis-open.org/archives/xliff/200207/msg00028.html, and wonder
if we could/should take this approach as a kind of guideline for gathering
further ideas for 1.1, or our general TC work. Here's a list of brainstormed
ideas about this:

1. We have some difficulties to close lists of values since we are unsure
   what's needed.

2. We may get a better picture of what's needed, if we either

   	2.1 look at particular examples for representing specific datatypes
          (e.g. Java property files) in XLIFF

 	or/and

	2.2 'get back to the source' (e.g. look at the XUL specification to get a
          list of all so-called 'common elements'); due to the TC members, we
           may have a good chance to do this

3. Presumably, it's beneficial if different tools generate the same XLIFF
   representation for a specific datatype (obviously, something like a 'canonical
   representation' which the different tool developer's could try to
   generate would be helpful for achieving this).

4. A repository of XLIFF examples for different datatypes may help people to
   overcome their fear of XLIFF ('Wow it's that easy to represent my Java stuff in XLIFF').

This mail is meant as feeler how you think about this.

The timing of the tasks and discussion with respect to building the repository of XLIFF sample
files from my point of view would be best if left until after we address the structural issues
that we are presently working on. The samples repository, with complementary narrative user
assistance, could be part of our 1.1 deliverables.

In case my explanations make sense to you, we could

a. add the work proposed to our project work and deliverables,  and push it onto the stack of
   existing tasks

b. already think about ways to approach 2.1 and 2.2 (e.g. ask a SUN
   representative for a proposal of how to represent Java resource bundles in XLIFF)

Best,
Christian