• HtmlTestMaterials

HTML Test Materials

HTMLAsSheAreSpoke is kinda messy; a big pile of tests will be at least as valuable as a spec for interoperability.

How about an HTMLTestJam March 10 and/or 11 along with SXSW Interactive?

testrunner.htm runs html5lib tests against browsers.

Note 31 July discussion of HTML WG test suite copyright policy.

Philip Taylor's Live Token Stream Viewer, described in his 28 July 2007 post to public-html.

See Anne van Kesteren 14 March 2007 post to public-html for pointers to some more test materials; Graham replies with a pointer to some encoding tests.

In a 20 March 2007 post to public-html, Lachlan Hunt takes issue with a number of XHTML 2 features; a reply from Holst makes a number of compatibility claims. It would be useful to see these substantiated by test cases. mobile test results by Simon Pieters seems relevant.

Test Suite: User Agent Support for HTML 4.01

• accesskey attribute tests

• HTML/TableAccessibility has some test materials to ascertain the state of support for table markup.

  • Table Accessibility Test Results

• Comparison of TABLE markup alternatives (based on CSS2, Section 17 example)

  • TABLE using id/headers and axis

  • TABLE using scope, axis, and speak-headers

• Comparison of TABLE markup alternatives to orient user to author's tab-navigation reflow using tabindex for an embedded FORM (based on Techniques for WCAG 2.0 example):

  • FORM in TABLE using scope alone

  • FORM in TABLE using the "title" attribute defined for each FORM control

  • FORM in TABLE using and headers/id

  • FORM in TABLE uses axis, and headers/id

  • FORM in TABLE using headers/id and scope

• Tests of the "cite" attribute defined in HTML 4.01 for Q and BLOCKQUOTE

  • Support for cite & known implementations

• Tests for Exposing Longdesc Targets

  • Known Longdesc Implementations

Tests to confirm these are claims about interoperability problems are most welcome:

• IE maps unknown start and end tags to empty elements; Other browsers assume unknown start tags are for non-empty elements with an inline content model

• attributes do not need enclosing quote marks; the XML /> syntax for empty elements is ignored even if attributes are enclosed in quote marks

• Gecko coerces unknown tags to upper case, Opera doesn't

• browser varations in treatment of XML namespaces with DOM-based work arounds using scripts

• IE will in some cases parse to a lattice not a tree

• behavior for multiple definitions of same ID value (only interesting in safari vs. firefox)

• The http://syntax.whattf.org/ schema project includes some above-parser-level test cases.

misc canvas test stuff to look at and summarize...

• Opera's Canvas test pages

• LazyIlluminati tests

a test from http://simon.html5.org/test/html/ was cited in a 9 Jul 2007 msg to the WG

various ECMAScript binding tests

-- Internationalization tests by Ishida and the I18N Core WG

-- numbers/microsyntax tests by Geoffrey Sneddon

Content Type Sniffing Tests

http://hixie.ch/tests/adhoc/http/content-type/ ; see also discussion in www-tag Aug 2006

Test frameworks

Mozilla automated testing (DanC hopes to study it further)

hmm... sounds like WebKit has automated testing stuff that's at least as mature. pointers welcome.