RE: Canvas Test Submission approval/feedback request

The non-rendering specific cases will fit into the test runner (just like the getElementsByClassName tests) and one will not have to click pass/fail.

-----Original Message-----
From: James Graham [mailto:jgraham@opera.com] 
Sent: Tuesday, October 05, 2010 2:34 AM
To: Anne van Kesteren
Cc: 'public-html-testsuite@w3.org' (public-html-testsuite@w3.org); Philip Taylor <pjt47@cam.ac.uk> (pjt47@cam.ac.uk); Kris Krueger
Subject: Re: Canvas Test Submission approval/feedback request

On 10/05/2010 10:22 AM, Anne van Kesteren wrote:
> On Tue, 05 Oct 2010 01:29:47 +0200, Kris Krueger <krisk@microsoft.com>
>> http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_pixe

>> lManipulation_imagedata_002.htm
>>
>
> I don't think assigning to readonly should throw per Web IDL.

Indeed; see section 4.5.4 [1]

> As for the others, are they not covered by Philip's test suite? That 
> seems a much better place as that is also automated whereas these are not.

Indeed. It is important to us, and will eventually be important to the progress of the spec, that tests are suitable for simple automation. 
Visual tests are not suitable for such automation. Given the problems that such tests have caused in trying to get CSS 2.1 IRs and the problems they cause in our test systems, we don't feel that we can approve such tests unless they come with special justification as to why other test types (javascript tests, reftests) could not be used instead.

With that in mind:

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_text_font_002.htm


Not automated, and no justification given.

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_text_fillText_001.htm


Not automated, and no justification given.

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_pixelManipulation_imagedata_001.htm


404

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_pixelManipulation_imagedata_002.htm


Incorrect, as discussed.

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_images_drawImage_001.htm


Not automated, and no justification given.

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_colorsAndStyles_createPattern_001.htm


I think this is OK. Seems to be the same as 2d.pattern.repeat.undefined from Philip's testsuite.

http://test.w3.org/html/tests/submission/Microsoft/canvas/canvas_canvasElement_001.htm


I think this is OK. Seems to be the same as toDataURL.zerosize from Philip's testsuite.


In summary, there are two tests here that are correct and suitable for 
the testsuite, but duplicate existing tests. Therefore I don't think 
they should be approved; keeping duplicates out in general will be 
difficult, but we should try not to introduce gratuitous redundancy.

As an aside, is there a plan to use the testharness.js API for writing 
tests? I thought we were trying to converge on a common API for as many 
tests as possible?

[1] http://dev.w3.org/2006/webapi/WebIDL/#put

Received on Tuesday, 5 October 2010 15:00:36 UTC