RE: Canvas tests

I took a peak at the first three tests (fallback http://philip.html5.org/tests/canvas/suite/tests/index.fallback.html).
The tests look to be a good start, though looking at the test.js file it appears to contain special logic for various browsers.
For example see function _getPixel(canvas, x,y)

// Work around getImageData bugs, since we want the other tests to
// carry on working as well as possible
 if (! _getImageDataCalibrated)
....
// Firefox returns premultiplied alpha
if (data2[1] > 100 && data2[1] < 150)
 _getImageDataIsPremul = true;
else
 _getImageDataIsPremul = false;
// Opera Mini 4 Beta returns BGRA instead of RGBA
....
try { ctx = canvas.getContext('opera-2dgame'); } catch (e) { /* Firefox throws */ }

It appears that this was needed to get the tests to run on more browsers.  Though I'd rather see tests run the same in all browsers, which would mean to remove these workarounds.

Thanks!

-----Original Message-----
From: Kris Krueger 
Sent: Monday, March 01, 2010 6:54 AM
To: Philip Taylor; 'public-html-testsuite@w3.org'
Subject: RE: Canvas tests

I can help review these cases and get them submitted to either cvs or mecurial.
Maybe you can review some of the cases I have submitted recently as well?

-Thx

-----Original Message-----
From: public-html-testsuite-request@w3.org [mailto:public-html-testsuite-request@w3.org] On Behalf Of Philip Taylor
Sent: Monday, March 01, 2010 5:03 AM
To: 'public-html-testsuite@w3.org'
Subject: Canvas tests

I've recently updated my old canvas test cases at 
<http://philip.html5.org/tests/canvas/suite/tests/> so that they better 
match the current version of the spec. There's also a results table at 
<http://philip.html5.org/tests/canvas/suite/tests/results.html> showing 
the current status of some browsers.

The aim has been to test all of the spec's testable requirements and 
various potentially-interesting edge cases - there's an annotated spec 
at <http://philip.html5.org/tests/canvas/suite/tests/spec.html> which 
shows the relevant test cases for most statements. It's not complete yet 
(there are large issues like drawing <video> which I haven't even looked 
at, and probably lots of minor bits missing), and there are probably 
lots of minor bugs too, but I think it's good enough for now and I'm not 
planning to work on it much in the near future.

(See <http://lists.w3.org/Archives/Public/public-html/2009Sep/0877.html> 
for some notes on the technical implementation of the tests.)

In the long term, an 'official' HTML5 test suite would have to include 
canvas tests, so this is probably a useful starting point. In the short 
term it'd be good to get more review of these test cases, and to have 
nicer mechanisms for running tests and reporting results, and to 
integrate them with browser developers' regression test suites. Are 
these things that the Testing TF could potentially help with?

-- 
Philip Taylor
pjt47@cam.ac.uk

Received on Tuesday, 9 March 2010 02:10:55 UTC