Integrate testsuite with web-platform-tests

[jgraham]

It seems like it would be a good idea to integrate this testsuite with web-platform-tests directly so that vendors who import that testsuite would also get these tests. Is there some reason that isn't sensible?

[fred-wang]

Yes, that's the long term plan.

[fred-wang]

So the only reason why the tests are not supported is that they are based on the draft http://www.mathml-association.org/MathMLinHTML5/ which extends the official MathML recommendation with some rendering rules based on TeX / OpenType MATH. In the long-term this can be proposed as a W3C note or something to the Math WG and so it would make sense to have it in the official W3C web-platform-tests. At least, the repo is already organized to make that import easy.

[fred-wang]

Update on this: Many tests have now been imported into the W3C test suite and there are three pending pull requests.

Remaining tests to handle:

1. http://tests.mathml-association.org/mathml/relations/html5-tree/entities-1.html
   This test uses a big htmlmathml.json file downloaded from https://www.w3.org/2003/entities/2007/htmlmathml.json using the tools/html5entities.py script. This test has not been imported into WebKit, I don't remember why maybe loading the htmlmathml.json file is causing some timeout. Perhaps this test should be rewritten and split.
   There were also the issue that the XML and HTML5 entities were different, I have to check the status of this.
   @jgraham Do you know whether there w3c/web-platform-tests already contains a test to check these entities? If so, maybe that one can just be removed.

2. http://tests.mathml-association.org/mathml/relations/css-styling/mathvariant-transforms.html
   This one is also generated using a JSON file and loads a big WOFF font, both generated from via mathml/tools/mathvariant-transforms.py. I think there was timeout issue on WebKit, so maybe it should be split.
   Also, for people using the web-platform-tests runner, it is not really easy to check at the naked eye. Maybe the glyphs should look more like real unicode code points to make them readable and distinguishable.

3. http://tests.mathml-association.org/mathml/relations/html5-tree/integration-point-1.html
   http://tests.mathml-association.org/mathml/relations/html5-tree/integration-point-2.html
   http://tests.mathml-association.org/mathml/relations/html5-tree/integration-point-3.html
   These tests pass on WebKit but not on Gecko, so I have to check if I'm not making any assumption on the default style of HTML elements.

[fred-wang]

3. These tests pass on WebKit but not on Gecko, so I have to check if I'm not making any assumption on the default style of HTML elements.

OK, these were only due to some font-family difference on the math tag. I opened web-platform-tests/wpt#4010

[fred-wang]

Latest update on this:

There are only two remaining pull requests to review and merge:
web-platform-tests/wpt#4010
web-platform-tests/wpt#4021

I'm going to remove test mathml/relations/html5-tree/entities-1.html, since these entities are already checked by html/syntax/parsing/named-character-references.html