RE: re p:filter-002

> I think that p:filter-002 should now throw an error and should be
> moved (if a test does not already exist) ... I think that calabash for
> example now outputs an empty sequence

And Calumet does, too. I think the test is correct, because the intent
*was* to have p:filter that matches nothing, and therefore returns an
empty sequence.

If the test said something like:

<p:filter name="step1" select="//p/text()"/>

then I would agree with you that it should throw an error. But as it
stands now, I think the test is correct.

Regards,
Vojtech