Re: [DOM TS Design: Test definitons]

>I'd say it's Perl-specific at this point.

Yep.

The DOM test suite will test for compliance against the DOM spec. If you
fail tests, you aren't compliant. If you deliberately don't implement all
the DOM behaviors, you should be able to predict exactly what tests you're
going to fail. (Note that passing a test suite never proves compliance
since it's impossible to test every possible case, but failing it --
assuming the test is correct -- definitely establishes a point of
noncompliance.)

If you want to discuss whether the DOM might define specific subsets and
compliance to those subsets, that's a topic for the IG list rather than
this one. Start by reading the open issues list (level_3_issues.html),
specifically the section titled "Lightweight DOM"; we _have_ considered the
issue before but it's unlikely to fly unless there is a subset whose exact
boundaries are widely agreed upon (meaning cross-langage and
cross-implementation) and which defines a fairly sharp functional/semantic
boundary. For example, the "read-only DOM tree" proposal passes the second
test but so far has failed the first despite several attempts.

______________________________________
Joe Kesselman  / IBM Research

Received on Friday, 23 March 2001 11:18:27 UTC