[XHR] Review of Microsoft's XHR tests

Hi,
I've just looked through Microsoft's submissions in 
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/
Though the scope is limited the tests are excellent and complement the tests Opera submitted quite nicely. My first question: can I just move those out of the /tests/submissions/Microsoft/ directory into the main /XMLHttpRequest directory?


Secondly, there is a small issue with the abort() tests. All those tests do a somewhat convoluted workaround for the fact that some XHR implementations (IE among them) fire readystatechange twice when readyState is 1 (I believe it's once for open() and once when send() starts). 
The spec was simplified in this commit, in response to some list discussion and feedback saying that the double event was not required for compat: https://github.com/whatwg/xhr/commit/15e86b74b03b9bb80186d6edfa8be091d2312522 


Hence, these tests are basically wrong per the spec as it is:


http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/abort-event-abort.htm - depends on open() send() sending two readystatechange events with readyState 1, times out
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/abort-event-loadend.htm - ditto, test will pass but won't actually test what it intends
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/abort-event-order.htm - ditto
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/abort-upload-event-abort.htm - ditto
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/tests/submissions/Microsoft/abort-upload-event-loadend.htm - ditto. Fake pass.


It's somewhat tempting to just do the small rewrite, but maybe WG members from MS wants to review this stuff a bit?
-- 
Hallvord R. M. Steen
Core tester, Opera Software

Received on Tuesday, 14 May 2013 16:35:42 UTC