A Web Standard Quickly: W3C Navigation Timing reaches Candidate Recommendation
Sometimes Web standards can converge quickly. The W3C standard for Navigation Timing started about six months ago in the newly chartered Web Performance Working Group. The working group’s objective is to provide interoperable features that enable developers to measure and improve performance of their Web applications.
Navigation Timing is the first specification from this this W3C working group, and in only six months went from a Working Draft to Candidate Recommendation. The community and representatives of Microsoft, Google, Mozilla, Opera, and Facebook worked together in an agile way to deliver a standard that works in IE9 and Chrome and will likely come to other browsers soon.
Is this standard “Ready and Done”?
The status Candidate Recommendation (CR) means that there is consensus agreement among participants at the W3C that the standard is “is stable and appropriate for implementation.” The W3C Directors can then call for implementations. The next step after the call for implementations is Proposed Recommendation, the final endorsement by the W3C.
Internet Explorer 9 Release Candidate and Chrome 10 Beta both offer vendor-neutral implementations of Navigation Timing. This means that Navigation Timing is an interoperable API and is ready for use by Web-developers today. The third IE9 Platform Preview was the first implementation of Navigation Timing, and IE9 was the first full browser to implement the specification with a vendor prefix. While transitioning from Last Call to Candidate Recommendation in the W3C, IE9 and Chrome removed the prefix. Now, Web-developers have access to performance information about their Web sites in an interoperable way.
For Developers and Privacy Enthusiasts
In the Release Candidate build of Internet Explorer 9, we removed the vendor prefix and moved towards the namespace window.performance. Underneath the performance namespace you’ll find two interfaces, navigation and timing. Navigation focuses on describing the type of browsing and browsing activity such as the number of server-side redirections that occurred while retrieving the root document, while the timing interface contains the time taken to fetch and load the root document.
To protect consumers from malicious sites, Internet Explorer 9 follows the specification and zeros out the timings for third-party sites that an attacker may attempt to use to determine your browsing history. Analyzing which third-party sites have a faster cross-site access time could reveal cache information.
What’s Ahead
With two implementations available, moving towards Proposed Recommendation should not take very long. You can see the Navigation Timing interface in action on the IE9 Test Drive. Submit your feedback so the working group can make sure we have designed and implemented an interface that meets your needs.
Special thanks go to Arvind Jain, Steve Souders, Zhiheng Wang, James Simonsen, and Tony Gentilcore from Google, Jonas Sicking from Mozilla, Sigbjørn Vik from Opera, and Jason Sobel from Facebook for their contributions to this work.
The working group has now turned our attention toward Resource Timing and User Timing, capabilities that enable developers to measure the performance of individual site resources and script. We expect these specifications to follow a similar implementation path over the next year.
—Jason Weber, Performance Lead, Internet Explorer
Comments
Anonymous
March 18, 2011
Thanks for the update. How about bringing about some agility in XML rendering in IE9? It appears to me, msxml engine works extremely slow in IE9 as compared to IE8.Anonymous
March 18, 2011
I found a cockroach in IE9. Steps to reproduce: > Open Internet Options and under the homepage. If you have bing.com in the list, remove it. > Open bing.com and click "Make bing my home page". When the popup appears, select the second option "Add this webpage to your home page tabs" > Refresh the page and you will find the "Make bing my home page " button again. If you had clicked the first option on the popup, this button won’t have appeared. > Now, click the button and select the second option again. Open Internet options, you will find the duplicate entry for bing.com. This is a clumsy bug. Please fix it. -- IE with spellchecker would be much more better than without it !!Anonymous
March 18, 2011
Please make the HTML/CSS editing more interactive, flexible and real-time like firebug ! Bundle of thanks if you listen to this one :)3)Anonymous
March 18, 2011
I think this goes to show that when browser vendors and developers are wiling to all work together, the web can move forward and progress in months, not years. Well done to everyone involved.Anonymous
March 18, 2011
The comment has been removedAnonymous
March 18, 2011
I enjoyed reading this post, It made me interested more to the given topic. Will wait for more posts! <a href="aqarmakkah.wordpress.com/.../a>Anonymous
March 18, 2011
@Juliano Nunes, MS has already explained that they would not incorporate some of the standards that Acid3 test verifies. Even the 100% result doesn't show how "well" the standard is implemented owing to the fact Acid3 is not a very comprehensive test doing in-depth/how-well analysis. There are number of benchmarks in which MS surpasses other browsers.Anonymous
March 18, 2011
Following are the scores; I have compiled from various benchmarks: PC = dell inspiron 1564 (corei3, 4GB RAM) with windOS7x64 MAC = MacAir with OSX 10.6.6 test262.ecmascript.org/default.html IE9 fails only 18 out of 10456 tests <-- which is the best thus far compared to any browser, any version till date sputnik.googlelabs.com/run IE9 fails only 66 out of 5246 tests <-- which is the best thus far compared to any browser, any version till date www.webkit.org/.../sunspider.html IE9 on PC: 549.4ms +/- 2.3% <-- compared to Chrome 10.8.63beta on Mac: 2832.1ms +/- 23.2%, Safari 5 on same PC: 677.3ms +/- 10.1%, FF4RC1 on same PC: 756.5ms +/- 4.6%Anonymous
March 18, 2011
krakenbenchmark.mozilla.org/index.html IE9 on PC: 25694.2ms +/- 1.0% <-- Chrome 10.8.63beta on Mac: ???, Safari on same PC: 31504.2ms +/- 11.2%, FF4RC1 on same PC: 11379.8ms +/- 3.6%, but far better than IE8 !! v8.googlecode.com/.../run.html IE9 on PC scores: 662 <-- Chrome 10.8.63beta on MacAir OSX10.6.6: 1721, Safari on PC: 997, FF4RC1 on PC: 886, but far better than IE8 !! Finally, MS's GPU-accelarated-graphics test ie.microsoft.com/.../Default.html IE9 on PC: 1784 rev/s <--- Chrome 10.8.63beta on MacAir OSX10.6.6: 0 (ZERO!!!) rev/s (failed!!), Safari on PC: 12 rev/s, FF4RC1 on PC: 1764 rev/s Also, check out Also, checkout test.csswg.org/.../results for CSS2.1 standards support in various browsers.Anonymous
March 18, 2011
@DanglingPointer: Is there a summary of those results (i.e. browser X passes Y tests and fails Z tests) ?Anonymous
March 18, 2011
@danglingpointer It seems like mozilla's Kraken test uses things like WebGL typed array elements to improve performance. Only browsers that support those WebGL element can use those and as such will perform better on Kraken.Anonymous
March 18, 2011
@hAl: IIRC, the whole point of Kraken is to test TraceMonkey's trace trees (or whatever those are called). Kraken should not be used to compare browsers, just like saying Chrome is better because it performs well on V8 Benchmark Suite is not a good conclusion.Anonymous
March 18, 2011
Great work on IE9! What I wonder is.. why not bring out updates for the development tools (F12) seperately. So we don't need to wait until IE10, but still can benefit from updates on the development tools.Anonymous
March 19, 2011
@juliano the 5 points missing on the Acid3 test are for SVG fonts and filters that the SVG working group will either retire or combine with other parts of the standard in the near future, so adding them to pass an arbitrary test has little point (and is hardly a minor update). the 'HTML 5' test you reference is also somewhat unbalanced in the weightings it uses and the state of the parts of HTML5 that it covers. The best test results are those done with the W3C test sets; there are links to the pass percentages of informal testing on the IE Testdrive site and the W3C is compiling lists of approved tests (the informal tests were run by Microsoft with other browsers, it will obviously be better for that to be done by the test working group independantly).Anonymous
March 19, 2011
I can't login to imageshack.us using either mode - standards mode or compatibility mode. Such is the problem with MANY sites. And when I go to download from the MS Download Center, the downloads are blocked as the page which starts the download is a popup. Before I can do anything with the notification bar to allow the popups, the page redirects! What an irony! Can't download from Microsoft.com using IE9.Anonymous
March 19, 2011
I have downloaded IE9 and installed with no problems on windows 7 home premium 64bit. I have enabled the status bar at the bottom of IE but it does not show the green progress bars when loading web pages,both 64bit and 32bit versions. I have googled this but the only hits I get refer to the status bar not the progress bars. I don't believe the progress bar has been done away with. It seems although there is the option of turning the status bar back on, it now is useless and does nothing. I noticed the same issue, the status bar also gets used by some addons like simple adblocker which is effectively useless on IE9 because of this. Personally I think the gui changes in IE9 are a joke, its almost an enforced full screen mode. IE9 is very fast but I felt the GUI changes made it unfunctional so I rolled back to IE8. what has baffled me is how many people have simply accepted major GUI changes as if it is no hinderence. I have moaned about it in a few places and got shot down by rude people telling me to move with the times. Seems many people now days assume newer always means better. sadgoonerAnonymous
March 20, 2011
IE team, I cannot view GIF animations at full speed when the image is zoomed. On a Windows XP computer with XP Picture & Fax Viewer, an image such as this one: upload.wikimedia.org/.../Prince_of_Persia_%281989_video_game%29_IBM_PC_Version_gameplay.gif plays smoothly even when zoomed. In IE9, there is a horrible lag. Other browsers do not have this lag. With Windows Photo Gallery no longer supporting animated GIFs, the ability to render them correctly rest with IE.Anonymous
March 20, 2011
@HW acceleration shame: I have it at 1000% zoom and it runs just as smoothly. It must be something wrong with your test machine. I also have it at 10% as well and it runs as fast as it does normally as well.Anonymous
March 20, 2011
There is horrible lag on that animatied gif if your computer runs in software rendering mode.Anonymous
March 20, 2011
@ajo, are you referreing to the updates like these; connect.microsoft.com/.../extensibility-in-f12-developer-tools . @hAl, it would be great if MEIS9 improve performance in those modules which V8 and Kraken are focusing. But overall, IE9 makes me feel good as compared to any other browser due to GPU driven graphics and exceptionally agile JScript engine. @ieblog, please update on the memory leak issues with which wikipedia is plaguing people on its IE9 page!Anonymous
March 20, 2011
@Danglingpointer From: blogs.msdn.com/.../ie9-final-release-rtw-minor-changes-list.aspx "Memory leaks in Sidebar gadgets have been mitigated."Anonymous
March 20, 2011
@hardware_acceleration guy: IE 9 doesn't run on XP so it seems unlikely that's your issue?Anonymous
March 20, 2011
@hAl, no wonder why WP is setting a bad example of spreading information and good example for unreliable information. Thanks for the quick reference! :-)Anonymous
March 20, 2011
I did not say I am running IE9 on XP. I am saying compared to any other browser or the Picture Viewer on XP, GIFs animations in IE9 with HW acceleration on Vista/7 play slower when zoomed. Reading comprehension fail..Anonymous
March 20, 2011
@HW acceleration shame: Well I seem to have it running perfectly fine on my intel i3 graphics card.Anonymous
March 20, 2011
"The working group has now turned our attention toward Resource Timing and User Timing, capabilities that enable developers to measure the performance of individual site resources and script. We expect these specifications to follow a similar implementation path over the next year." So that means IE10 (or IE9.1?) will be out in roughly one year? Great news.
