From Karl's questions:

- What's the difference between Description and Purpose?
In the XSLT test system, "Purpose" is limited to 255 characters or less, with no internal new-lines. It can then be used for log entries, tables in HTML, etc. with an assurance that it won't be overly large.

- What's the content of Expected Results?
In some test disciplines, this could be a direct representation of the desired value, but most would put a pointer to a file that contains the expected result. Thus, you want some meta-data about type and whether the content is a filename (or file path).

- What are the different values of Status
As I recall the last discussion on this point, the idea was that QAWG would dictate a minimum set of statuses, but that the WG could expand the set to fit their protocol. For example, a protocol where tests are written first might have a status code to indicate that the test shows intended behavior on a particular point, but that the specs had not yet been written around that point and no consistency check with all other points had been performed. QAWG would require status codes that signify that a final disposition had occurred, like Accepted and Rejected, to ensure some clarity and uniformity when exposing tests to the world.

- Why there's no property for modified ? or should a date be attached
to the node Status, with also possible contributor associated to this
node?
Actually, the idea of a date attached to status is interesting for various reasons. I take that to mean the date that the current status was assigned to the test.

Also, I think that HowToRun is probably too thin, and that the "scenario" mechanism I have proposed earlier will be needed to represent a bundle of expectations about setting up to run, running, evaluating the results, and cleaning up.
.................David Marston