Re: Review of the test failures

Hi Boris,

On 06/01/2015 18:34 , Boris Zbarsky wrote:
> On 1/6/15 11:32 AM, Robin Berjon wrote:
>> http://w3c-test.org/dom/nodes/MutationObserver-attributes.html
>>    • Implementations don't properly default attribute: true.
>
> Which implementations?  This test seems to pass in Gecko.

Sorry, I should have been clearer here, and notably included a link:

     http://w3c.github.io/test-results/dom/less-than-2.html
     http://w3c.github.io/test-results/dom/all.html

The first one has the failures that are of most interest (since they 
fail in most places), the latter has all the data (but is long).

When I wrote "implementations", it is to mean "several implementations". 
I wasn't interested in things that don't fail much since that tends to 
indicate that things are headed in the right direction and there is 
mostly QoI clean-up.

To give an expanded take on that labelling, it essentially means: "On 
this point the specification makes sense (to me), the test accurately 
exercises it, if you're one of the fine folks failing here and you think 
you have a good reason to fail it would be particularly useful to speak up."

Gecko does pass a fair bit more tests than others. I would say that of 
particular interest where Gecko is concerned would be to look at the 
Bugzilla entries I listed and weigh in on a specific direction.

>> http://w3c-test.org/dom/errors/exceptions.html
>>    • IDL stuff
>
> That link is 404, so hard to comment on.

Yes, work continues on the suite and some stuff has moved or been 
deleted, sorry.

>> http://w3c-test.org/interfaces.html
>>    • the usual WebIDL stuff
>
> Again 404.  Did you mean http://w3c-test.org/dom/interfaces.html ?

Yes. That (and all the 404s that don't have /dom/ in the path) is the 
result of my losing a fight with Thunderbird's find&replace. Sorry about 
that.

> which per current WebIDL spec should not throw, but that's not very
> compatible with ES6 subclassing so in practice UAs throw and this needs
> to be fixed in the web IDL spec and in this test.

Yes, I'm not sure it's worth looking closely at anything IDL-related 
here. WebIDL produces failures in every single API we have :)

-- 
Robin Berjon - http://berjon.com/ - @robinberjon