RFE: evaluate test results by graph match rather than text diff

We have occasionally lamented the fact that the cwm/swap
tests use diff to evaluate test results, when most of them
really should be using graph matching.

We haven't had any particular reason to fix it, yet, so we
haven't.

I recently learned that the diff-based tests are causing
difficulty for some collaborators; i.e. the folks
working on pychinko.

I suspect we could share more tests with Euler too,
if we fixed our test harness.


 http://svn.mindswap.org/pychinko/

http://www.ilrt.bris.ac.uk/discovery/chatlogs/foaf/2004-08-20#T10-17-15
yarden katz: pychinko: Rete-based CWM clone

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/