RE: Failures in Canvas2D test suite

> Looking deeper at one of those [1], the 2d context specification seems clear to me:
> [[
> If the image argument is an HTMLImageElement object that is not fully decodable, or if the image argument is an HTMLVideoElement object whose 
> readyState attribute is either HAVE_NOTHING or HAVE_METADATA, then the implementation must return null.
> ]]

My point was that how images are loaded, is not specified by canvas.  The test is not failing because of a canvas feature.

> would be careful here as well. Some of those are well defined in Web IDL
That's true. There's no '?' in the IDL, so we're requiring an object.
I'm unsure if the normative text should still call that out thought..

> Just tried again on two firefox 26 and get the same failure
I was on FF 23 so maybe someone changed this.

-----Original Message-----
From: Philippe Le Hegaret [mailto:plh@w3.org] 
Sent: Monday, September 16, 2013 11:35 AM
To: Rik Cabanier
Cc: HTML Test Suite; eliotgra@microsoft.com; Jay Munro; Tom Wiltzius; Michael(tm) Smith
Subject: RE: Failures in Canvas2D test suite

On Mon, 2013-09-16 at 10:51 -0700, Rik Cabanier wrote:
> Hi Philip,
> 
> I quickly looked over the failed results.
> 
> For the first table, the tests are not really testing canvas pattern but the 'loading state' of images which is not part of canvas 2d.

Looking deeper at one of those [1], the 2d context specification seems clear to me:
[[
If the image argument is an HTMLImageElement object that is not fully decodable, or if the image argument is an HTMLVideoElement object whose readyState attribute is either HAVE_NOTHING or HAVE_METADATA, then the implementation must return null.
]]

Neither Firefox or IE do so. It may well be that we classify this one (and possibly others) as an implementation bug.

> For the second table:
> - the type of exception for the imageData failures are not in the spec 
> so this is undefined behavior
> - same for the gradient failures.
> - same for drawImage

I would be careful here as well. Some of those are well defined in Web IDL. However the 2d context spec says "Interfaces are defined in terms of Web IDL.", which could be interpreted as "following the IDL syntax and not pretending to follow the ECMAScript binding". A clarification in the spec might be in order here. Note that Wed IDL defines 3 levels of
conformance:
 http://www.w3.org/TR/WebIDL/#conformance


> For the third table, the 'font' property is supposed to return the 
> serialized value. This is outside the scope of the canvas 2d spec (and 
> also seems underunspecified)

Agreed.

> The test in table 4 is passing for me in Firefox 23 (but not ie10)

Just tried again on two firefox 26 and get the same failure. weird. Are we sure Firefox didn't change this recently?

Philippe

Received on Monday, 16 September 2013 20:07:24 UTC