PLS Issue R103-30 (i18n core comment 30)

Some input for todays discussion, see
http://lists.w3.org/Archives/Member/member-i18n-core/2006Dec/0013.html:

I'm trying to understand what we want to achieve here:

1) Should a PLS processor be able to "understand" xml:lang or the dir
attribute (i.e. do s.t. useful with them), or
2) Should an author of PLS documents just be able to use these attributes?
In the case of 2), a different application would make use of the
attribues, e.g. an HTML browser which gets (X)HTML, generated via XSLT
from the PLS source.

The PLS draft says in sec. 4.8:
"Because the examples are explicitly marked, automated tools can be used
for regression testing and for generation of pronunciation lexicon
documentation."
in other words: there is no specific processing requirement for the
(textual or complex, i.e. with markup) content of the <example> element.
Hence, I think we can't really push for 1).

Now, if we push for 2), the next question is:
2a) Do we want xml:lang and the dir attribute, and span element
explicitely in the PLS schema, or
2b) Do we just want to make sure that an PLS author does not get punished
by an validation error if he uses these attributes and / or span?

PLS uses RELAX NG as a schema language, which gives you the following way
to realize 2b):

element example {
  attribute * - (pls:*) { text }*,
  (text | anyElement)
}
anyElement =
  element * - (pls:*)  {
    (attribute * - (pls:*)  { text }
     | text
     | anyElement)*
  }

This reads as: the <example> element can contain any attribute except from
the pls namespace (expressed via "attribute * - (pls:*)"), and mixed
content. In the mixed content, there can be "text" or "anyElement", that
is: any element except from the pls namespace (expressed via "element * -
(pls:*)"), again with any attribute from the PLS namespace and mixed
content.

If we agree on 2 as the way to go, and the PLS folks refrain to do 2a), I
would propose 2b) .

Felix

Received on Tuesday, 12 December 2006 01:37:55 UTC