See also: IRC log
Norm: We may cancel it, but we'll leave it on the books. No regrets heard.
Norm: Is that something I do, or Henry, do you do it?
<scribe> ACTION: Henry to investigate the charter extension process [recorded in http://www.w3.org/2008/08/14-xproc-minutes.html#action01]
No comments heard.
<scribe> ACTION: Henry to update the XSD and DTD files then give Julian the green light to publish [recorded in http://www.w3.org/2008/08/14-xproc-minutes.html#action02]
Norm looks for help and advice in constructing an adequate test suite on the assumption we'll make it through Last Call and get to CR easily.
Norm: There are already about 80 or so at tests.xproc.org
Henry: What's the metadata you need?
Norm: There's documentation for
the test format at tests.xproc.org, but I'm happy to adjust it
... We will stall if we don't get a test suite...
Henry: My metadata question was
about coverage. We need to give some thought to the metadata to
describe *what features* are tested.
... We need to be able to characterize the tests to make the interoperability claim have some teeth.
Norm: I was going to analyze the test suite for missing steps and missing options on steps.
Henry: That's a good start, but
there are also questions about the topology of the pipeline and
the inheritence of variables and defaulting of pipes,
... Those aren't as easy to enumerate.
Norm: I was also hoping to have tests for every static and dynamic error as well, that covers many of those things.
Richard: The XML test suite doesn't attempt to identify the "correct" error and this has sometimes caused problems.
Henry: Having explicit error codes helps a little bit.
Norm: Metadata is a good idea, maybe we should have some email discussiona bout how to devise metadata that covers the spec.
Vojtech: Regarding error codes, right now we have gaps in the numbering. Is that ok?
Norm: I think it is. We talkeda bout it once before and decided that the cost was larger than the benefit.
Vojtech/Henry: I'm ok with that.