Re: RC4 Invalid tests

On Dec 13, 2010, at 3:37 PM, fantasai wrote:

> On 12/13/2010 11:55 AM, Peter Linss wrote:
>> The harness is currently reporting 78 tests from RC4 as invalid:
>> http://test.csswg.org/harness/results?s=CSS21_%HTML_RC4&f=23
>>
>> This means that these tests were reported as invalid (from a formal  
>> implementation report by a vendor) in previous versions of
>> the test suite and have not been changed since that report.
>>
>> I accept that it's possible that some of these tests have been  
>> examined and found to be valid after all, if that's the case, I
>> need a list of those so I can clear the invalid state. If they're  
>> genuinely invalid, then these are the tests that need fixing
>> first for RC5...
>
> The page breaking tests need a rehaul. That'll likely take a few days.
>
> anonymous-boxes-001 was fixed by reference:
>   http://lists.w3.org/Archives/Public/public-css-testsuite/2010Dec/0098.html

Ok, my diff process didn't account for diffs in reference pages. I  
fixed that and removed all grandfathered results for tests where the  
reference page changed (and marked the affected tests as 'changed').

I also found that I wasn't accounting for the case where neither the  
test nor the reference page changed, but a new reference page was  
added to a test. That's also fixed and grandfathered results are  
removed for the affected tests (and the tests are marked as 'changed').


> I believe table-valign-001 and table-valign-002 are valid per Issue  
> 26:
>   http://lists.w3.org/Archives/Public/public-css-testsuite/2010Oct/0314.html

Ok, removed 'invalid' results for those tests.

That still leaves us with 74 invalid tests in RC4 (and a handful of  
tests now marked as 'changed' that weren't before and are now missing  
results...). Hopefully these will all be addressed in RC5.

Peter

Received on Wednesday, 22 December 2010 23:35:41 UTC