* Sam Ruby <[EMAIL PROTECTED]> [2006-04-01 01:50]:
> It would be helpful if people were to update:
> http://intertwingly.net/wiki/pie/XmlBaseConformanceTests

For that matter, I’ve been meaning to address some weaknesses in
that test suite which Liferea 1.0 highlights. Liferea does URI
fixup for Atom links in its feed parser, but merely uses the
entry’s alternate URI as the base URI when rendering content.
So it succeeds legitimately on cases that test things like
atom:link, but then accidentally succeeds on a number of cases
that involve atom:content where it should be failing.

I’ve been meaning to add some aggressive tests which use xml:base
values that differ drastically from the nearby alternate URIs in
order to smoke out such coincidentally passing tests, as well as
some intentionally evil tests with `type="xhtml"` where xml:base
is set on elements inside the xhtml:div. I expect to see a lot of
aggregators fall from grace with such an expanded test suite.

Sam: is it possible to host the test suites directly on the wiki,
by having pages that consist entirely of verbatim text? Ideally,
the content should be rendered inside the wiki chrome using
`<pre>` tags, but be downloadable without the chome by way of
adding something like `?display=raw;type=application/atom+xml` to
the page URI. That would make it much easier for more people to
pitch in. I find the collection of tests we have so worryingly
minimal; a lot of the currently lesser used corners of the format
are not being tested at all. It makes me nervous that dirty data
based on current incomplete implementation behaviour may become
too widespread for aggregator developers to be able to ignore it.

Aristotle Pagaltzis // <http://plasmasturm.org/>

Reply via email to