> I agree that the set of metadata elements needs to be extensible, but I
> still don't think we need header to get this. For example, we can do what
> we do in RELAX NG and allow arbitrary namespaced qualified elements.
OK. That seems to be fine.
> I noticed you have a metadata element for saying whether it depends on a
> XML Schema namespaces. How about something like
>
> <requires datatypeLibrary="http://..."/>
>
> Any others?
Not off the top of my head. I guess we can add elements later if
necessary.
> Allowing resources both in <testCase>s and in <testSuite>s would make
> things rather complicated.
Umm, then I'd rather propose to pull out <resource> on the ground that
1. use of resources involves use of <externalRef>s and <include>s; it is
hard to write a generic (processor-dependent) test framework that
properly resolves those resources for the processor.
2. there probably are test cases that cannot be written by using the
test suite format anyway (e.g., a test case that uses DTD.)
I said that I'm using resources to share common file between test cases,
but it was not quite correct. I do have test cases that share a file
between test cases, but they are not in this suite format; they are just
a bunch of small files now.
How does this sound for you? Or <resource>s are too vital for you to
throw away?
regards,
--
Kohsuke KAWAGUCHI +1 650 786 0721
Sun Microsystems
