XInclude under-implemented tests

> * The features tested by Nist-include-xx, where xx is
>         19, 24, 25, 48, 49, 51, 52, 56
> and eduni-3 appear to be under-implemented. 

> ACTION to Richard:  Go through those listed and send email.

19 is not well-tested because it relies on the xpointer scheme.
But it is supposed to test inclusion of a single item, which is
tested in many other tests.

24 and 25 are even worse than they look.  They are negative tests,
which are supposed to produce errors.  MT "passes" because it gets an
unrelated error (because it only handles a subset of the xpointer
scheme).  They are supposed to fail because they produce documents
without any elements.

48 and 49 are tests of infosets with unexpanded entities.  Only certain
non-validating parsers can produce these, and neither MT nor XOM does.
I don't know why libxml fails 48.  Is it a problem if we say what
a certain class of processor must do, and there aren't enough of that
class of processor?  Wouldn't it be worse to leave it unspecified?

51 and 52 test the ability to include a comment and a PI.  This can't
be tested with any of the required xpointer schemes.  Only libxml
supports the xpointer scheme well enough to pass.

56 tests unparsed entities.  Neither MT nor XOM provide access to these.

eduni-3 tests xml:lang fixup.  XOM should pass this test and we hope
to get a result for it.

-- Richard

Received on Wednesday, 15 September 2004 16:39:51 UTC