Markup Validator's test suite - using EARL?

Hi.

I am currently working on a test suite for the Markup Validator, which 
will eventually be semi-automated through the use of the Log Validator.

Currently the "suite" [1] is just a page with links pointing to either 
validation pages (using the test instance of the validator) or other 
"test suites", with notes about whether the page should validate or 
not, sometimes a description of what is being tested (XML, charset, 
etc). I would like to clean this up, bind it all using n3 and interface 
it with the Log Validator.

[1] http://validator.w3.org/dev/tests/


After a better look at how MUTAT [2] works, I started wondering whether 
we could actually use EARL for the test descriptions.

[2] http://www.barbwired.com/nadiaweb/tester/ (we have a version at 
w3.org but it doesn't work as well as this one, need to contact Nadiah 
about it)


Indeed we have:
- web resources (our test cases)
	- which we know is invalid "specX" or not
	- which is used to test this or that part of the validator (XML, SGML, 
charset, ...)

- we know that for each of these test cases the current validator "gets 
it right" or not, which could certainly be expressed using EARL 
assertions.


Without really wanting to start (again) a discussion about a generic 
Test Case Description Language, I was wondering if anyone more familiar 
with EARL and/or Test Cases would have comments and ideas about this.

cheers,
-- 
olivier