Common pool of examples and validation

Just worked through another case where Firefox and IE behavior differ in
"interesting" ways.  Yet another reminder that without extensive validation
tests, implemented browser behavior will differ.  Without an healthy set of
examples, behavior described in the specification may be incomplete.

So I have an example.  Seems useful to compare against the HTML5
specification, and to check that validation tests cover desired behavior.
Yes, I know it is a little early.  On the other hand, it would be useful to
start collecting a common pool of examples - as encountered by developers -
so we later have a pool to draw on, both to insure completeness in the HTML5
specification, and to build validation tests.

Guess my point is we should start collecting examples as early as possible.