SVG Test Suite (was): SVG/Canvas DrawImage Issue

Wow, I didn't consider this helping the Test Suite. Nice observation.

Jeff: You also asked previously whether or not you had to continue to maintain the Test Suite conformance site; to that end I have some thoughts, and would love to hear reactions.

I promised at the previous face-to-face that I would have a detailed analysis of the SVG Test Suite 1.1 2nd edition.  This is still in progress and I hope to get a detailed report out soon.

I have two key issues that I am trying to address.

First is whether or not a test is working as expected; I believe I am still finding a handful of these and want to make sure they are fixed if they are indeed incorrect.

Second, I think we want to communicate clearly to developers that want to use SVG what the interoperable set is.   From the analysis based upon the current test suite, I don't believe this is clear.  My proposal will likely include modularized tests, perhaps based upon "hasFeature".  Right now, , for example I believe there are SMIL tests where I don't expect them.  From the title and the category of the test, a developer might presume that some browsers fail certain areas when for all intents and purposes they don't (e.g. pointer events, gradients, etc.).

I hope to have this proposal soon and I hope that this proposal will be readily accepted by the working group.

In terms of where the results should be hosted, or who should run them, I believe that while vendors can/should post their results, it would be great if we could all agree on each implementation's conformance, and deliver these to an independent site (or W3C if they are open to that)

Patrick Dengler


From: www-svg-request@w3.org [mailto:www-svg-request@w3.org] On Behalf Of Jeff Schiller
Sent: Tuesday, July 20, 2010 10:59 AM
To: Patrick Dengler
Cc: www-svg@w3.org
Subject: Re: SVG/Canvas DrawImage Issue

Awesome!  Can't wait to have drawImage(mySvgImage) in <canvas> :)

Opens the door to automating SVG test cases too - use canvas pixel operations to check the rendered results of a SVG file.  Of course that requires all browsers to support this :/

Intuitively, I guess I would have thought it was percentage of the <canvas> element's dimensions.

Whatever the decision might be, it looks like an update to http://dev.w3.org/html5/canvas-api/canvas-2d-api.html#images will be required.

Jeff
On Tue, Jul 20, 2010 at 6:51 AM, Patrick Dengler <patd@microsoft.com<mailto:patd@microsoft.com>> wrote:
>From the canvas spec, drawImage states that drawImage should draw based on the intrinsic dimensions of an image.

We need clarification on what to do if the intrinsic dimensions of the source are percentage based.

Webkit establishes the size the based on window size when you loaded it.  When you resize the window, you get different results on the Canvas.

Thoughts/Clarifications are welcomed.