Bug 40829 (html5test)

Summary: Get a perfect score on html5test.com
Product: WebKit Reporter: Trevor Downs <cyberskull@mac.com>
Component: WebKit Misc.Assignee: Nobody <webkit-unassigned@lists.webkit.org>
Status: NEW    
Severity: Enhancement CC: abarth@webkit.org, alexgtk19@gmail.com, arv@chromium.org, ddkilzer@webkit.org, donggwan.kim@samsung.com, eoconnor@apple.com, eric.carlson@apple.com, eric@webkit.org, grahamperrin@gmail.com, laszlo.gombos@webkit.org, peter@chromium.org, rakeshchaitan@gmail.com, Reelix@gmail.com, rodger.combs@gmail.com, sherlock@genome.stanford.edu, staikos@kde.org, syoichi@outlook.com, tonikitoo@webkit.org, webkit.org@boblet.net, webmaster@keryx.se
Priority: P2 Keywords: HTML5
Version: 528+ (Nightly build)   
Hardware: All   
OS: All   
URL: http://html5test.com/
Bug Depends on: 3251, 19264, 27247, 29326, 29359, 29363, 32624, 33938, 38157, 38994, 41123, 42903, 43572, 43668, 44358, 44515, 45950, 58454, 61140, 68156, 68609, 73177, 76760, 76766, 78244, 83533, 84353    
Bug Blocks:    

Description From 2010-06-18 05:13:11 PST
The website <http://html5test.com/> tests browsers for the existence of HTML5 features and can be used to gauge the progress of features. I'd like to see WebKit hit 300/300 before the other engines do. Note that this site only currently checks for the existence of features, not that they are correct.


Current scores (running on Mac OS X 10.5 PPC):

WebKit 533 (Safari 5): 208 + 9.
Firefox 3.6.3: 139 + 4.
Opera 10.53 (Presto 2.5.24): 129 +4.

WebKit r61351: 230 + 9.
Firefox 3.6.4 (Build 7): 139 + 4


Note that I have the Xiph.org Ogg/Theora/Vorbis plugin installed, giving my machine an extra point.
------- Comment #1 From 2010-06-18 05:21:42 PST -------
I forgot the iPhone!

iPhone 3G, iPhone OS 3.1.3 (iOS 3.1.3):
WebKit 528.18: 125 + 7
Opera Mini: 22.
------- Comment #2 From 2010-06-21 19:12:19 PST -------
iPhone 3G/iOS 4, WebKit 532.9: 185 + 7
------- Comment #3 From 2010-06-22 10:00:28 PST -------
I'm not sure if getting 300/300 on this test should necessarily be a goal. But it's interesting that this test claims there's no "HTML5 tokenizer" or "HTML5 tree building" in ToT.

The very first subtest fails: div.innerHTML = "<div<div>" is expected to add a node with name "DIV<DIV".
------- Comment #4 From 2010-06-24 06:21:00 PST -------
WebKit r61744: 231 + 9

What is ToT?
------- Comment #5 From 2010-07-01 10:00:49 PST -------
WebKit r62241 235/300 +9
------- Comment #6 From 2010-07-07 05:13:07 PST -------
WebKit r62632: 236 + 9
------- Comment #7 From 2010-07-09 12:34:30 PST -------
WebKit r62909: 232 + 9.
------- Comment #8 From 2010-07-21 11:10:40 PST -------
(In reply to comment #2)
> iPhone 3G/iOS 4, WebKit 532.9: 185 + 7

This is being tracked for iPhone using:  <rdar://problem/8215076>

We also have radars covering differences between Safari 5 and iOS 4.0.x, so there's no need to file other bugs.

We will also eventually pick up fixes committed to trunk as well.
------- Comment #9 From 2010-07-23 08:01:58 PST -------
WebKit r63958: 222 + 10
------- Comment #10 From 2010-07-29 06:10:02 PST -------
I’m all for progress, but after the fun with HTML5 form validation I’d prefer usable HTML implementations over rushing to ship:
https://bugs.webkit.org/show_bug.cgi?id=40747
https://bugs.webkit.org/show_bug.cgi?id=40761

That said, keep on truckin’ Webkit peeps :)
------- Comment #11 From 2010-08-17 00:18:36 PST -------
WebKit r65398: 227 + 10 (Tokenizer now detected!)
------- Comment #12 From 2010-08-18 08:35:05 PST -------
(In reply to comment #11)
> WebKit r65398: 227 + 10 (Tokenizer now detected!)

Why did the score go down from a high of 236 + 9 in Comment #6?
------- Comment #13 From 2010-08-19 03:41:54 PST -------
(In reply to comment #12)
> (In reply to comment #11)
> > WebKit r65398: 227 + 10 (Tokenizer now detected!)
> 
> Why did the score go down from a high of 236 + 9 in Comment #6?

Mainly due to WebGL being broken/gone. That is 10 points there.
------- Comment #14 From 2010-08-25 22:40:29 PST -------
Webkit r66052: 232+12 !
------- Comment #15 From 2010-11-08 00:42:41 PST -------
WebKit r71484: 233 + 12! <output> now supported & detected.
------- Comment #16 From 2011-01-06 01:28:36 PST -------
WebKit r74228: 235 + 12, I'm guessing figure and figcaption support.
------- Comment #17 From 2011-01-30 11:57:54 PST -------
May I request that we move the URL to http://beta.html5test.com/ and change the title to "Get 400/400 points on the HTML5 Test"?
------- Comment #18 From 2011-01-30 16:52:10 PST -------
I get 288 + 9 points, on r75891 at http://beta.html5test.com/.  Note, to enable WebGL, you have to do:

defaults write com.apple.Safari WebKitWebGLEnabled -bool YES
------- Comment #19 From 2011-01-30 17:41:59 PST -------
In normal, I get 245+15. In beta, I get 288+15. This is with WebGL plus every media plugin imaginable.
------- Comment #20 From 2011-01-30 19:52:44 PST -------
(In reply to comment #17)
> May I request that we move the URL to http://beta.html5test.com/ and change the title to "Get 400/400 points on the HTML5 Test"?

I think if any tests should be strived for, the stable release tests should be the priority over the beta tests. While 400/400 sounds better, the beta tests could be updated at any time completely hosing the score of anything being tested.
------- Comment #21 From 2011-01-30 19:54:57 PST -------
(In reply to comment #18)
> I get 288 + 9 points, on r75891 at http://beta.html5test.com/.  Note, to enable WebGL, you have to do:
> 
> defaults write com.apple.Safari WebKitWebGLEnabled -bool YES

Has WebGL been re-enabled? I have it turned on but none of the WebGL code runs.
------- Comment #22 From 2011-03-06 13:02:33 PST -------
The HTML5 test has been updated, it is now 400 points!

Here's how the browsers do (based on what I can run on 10.5 PPC ;) ):

* WebKit r80210: 273 + 12

* Safari 5.0.3: 228 + 9
* Safari (iOS 4.2.1): 206 + 7
* Firefox 3.6.14: 155 + 4
* Opera 10.63: 214 + 7
------- Comment #23 From 2011-04-13 23:16:33 PST -------
I would also like to see this.

In fact - I would be happy with a base score of 300 points - Just 9 more to go! :D
------- Comment #24 From 2011-04-13 23:52:11 PST -------
WebKit r83424: 271 + 15
I updated Perian which now supports WebM, so that accounts for the bonus improvement. I wonder what broke to lower the main score?

WebKit r83750: 271 + 15
------- Comment #25 From 2011-06-06 16:53:55 PST -------
WebKit r88189: 269 + 15
------- Comment #26 From 2011-06-06 23:52:22 PST -------
Chrome 11.0.696.77

293 and 13 bonus points

Best so far! :D
------- Comment #27 From 2011-07-14 05:07:20 PST -------
Is there a bug for http://www.html5accessibility.com/HTML5ASS/
------- Comment #28 From 2011-07-14 05:19:29 PST -------
(In reply to comment #27)
> Is there a bug for http://www.html5accessibility.com/HTML5ASS/

You should just create one.
------- Comment #29 From 2011-07-21 17:49:36 PST -------
Safari 5.0.6 (5533.22.3): 293 + 15.
------- Comment #30 From 2012-01-20 19:42:20 PST -------
I went through the test suite this afternoon and broke down all our failures in this spreadsheet:
https://docs.google.com/spreadsheet/ccc?key=0AlC4tS7Ao1fIdHZtUDNsOFNFYmxGbzZFVVJERUswZFE#gid=0

The end result is that we're already the "best" engine according to html5test.com, and almost all of the remaining tests we either already pass (and the suite is wrong) or we have code in progress (either behind compile-time flags, or with webkit- prefix guards).

About the only interesting implementation work left might be <command> (bug 58454), HTMLInputElement.dirname (bug 76766), and seamless iframes (bug 45950).
------- Comment #31 From 2012-01-22 04:21:18 PST -------
As of the 2012-01-01 update:
* Safari 5.0.6 (5533.22.3) on 10.5.8 PPC: 302 + 15
------- Comment #32 From 2012-11-12 13:56:43 PST -------
WebKit r134216 (OS X 10.8.2 Intel 64): 388+15
Safari 6.0.2 (OS X 10.8.2 Intel 64): 378+15