Up until yesterday we had no way to spot a failed test. I'm referring not to a failure such as an HTML validation test finding errors but a test that fails to run at all.
Some full-site HTML validation tests were stalling on a small number of tests. In particular there was a ~150 page site where the HTML validation tests for 3 URLs were never completing.
It was clear, after running the relevant URLs manually through the HTML validator, that these were pages that could not be validated.
The URLs for which test were never completing were pages were advertising
themselves as being
utf-8 encoded but which included
character references that did not map to Unicode; not something I'd
Tests that cannot run are now picked up as failures. So far this covers cases where the HTML validator cannot perform any validation.
When running a test you can now see another bar below those that list the number of queued, in-progress and completed tests to show the number of tests that failed.
Once the full-site test has finished (or has been cancelled if you don't fancy waiting), the results page for a failed test will detail why the test could not run.