Re: Using fundamental/extended as an organization principle (was isSu pported12)

> The distinction between fundamental and extended is less significant than
the difference between tests that depend on parsing XML and those that
depend on parsing HTML.
>
> However, I think it is unlikely that we could write a significant body of
tests for DOM core that would work on both an HTML and XML implementation.
So currently we have a body of tests for the DOM
> L1 and L2 Core that pre-suppose the implementation can read XML.  So maybe
we should have put the tests that depend on XML support in
tests/level-1/core/xml and put html tests in
> tests/level-1/core/html to avoid implying superiority over html.  And if
we were able to write tests that could work on both we could put them in
tests/level1/core.
>
[mb] This discussion has to do with what DOM modules should be defined, and
where, not XML vs HTML.  Right now, we have all of our tests under core.
Since core must be implemented by everyone, we are implying that all should
support all of the tests.  In fact, the extended interfaces are not part of
the
"Core Module", and do not have to be implemented by anyone who does not
support the "XML Module".  This is somewhat confusing, given that both
modules are defined in the DOM Level-2 Core spec.  I would expect that some
of the errors you are finding in the Mozilla implementation are really
failures to
support the extended interfaces (XML Module), and fall into this category.

> However, I'd like to hold off discussing any directory structure change
until we have had a chance to get the L2 stuff running.
>

[mb] It's not simply a discussion of directory structure, and has to do more
with analyzing the results of the tests you are currently running.

> Can you give us the status of the HTML DOM tests?
>
[mb] The HTML DOM tests are next in the queue after the DOM Level 2 tests
are running.  They will have to be rewritten, as they
presupposed the ability to load an XML file synchronously, and attempted to
make use of common functions to get at nodes, and
their children.  I also expect that we will break out the one large HTML
file into sections, and have a set of tests correspond to a
smaller HTML file created soley for that purpose.  They will be committed as
each section is completed.

--Mary

Received on Thursday, 25 October 2001 21:34:37 UTC