HTML5, Native: Third IE9 Platform Preview Available for Developers
As developers experiment and begin the transition from writing today’s websites to building HTML5 applications, they will test the limits of browsers. For example, the huge difference between hardware accelerated HTML5 video and plain HTML5 video support in a browser was clear in the demo we showed at MIX.
Because some browsers run on many different operating systems, there can be a tendency to use a “least common denominator” approach to implementing HTML5. By using more of the underlying operating system, and taking advantage of the power of the whole PC, IE9 enables developers to do more with HTML5. Running through Windows, instead of just on Windows, makes a big difference; the web runs more like a native application. This is consistent with our approach of architecting HTML5 support in, from the ground up, rather than just grafting in some HTML5 features.
The third Platform Preview of Internet Explorer 9, available now, continues the deep work around hardware acceleration to enable the same standards-based markup to run faster. This is the latest installment of the rhythm we started in March, delivering platform preview releases approximately every eight weeks and listening to developers. You’ll see more performance, same markup, and hardware accelerated HTML5.
This video of same markup running in IE9 and other browsers shows what hardware acceleration means for the new, graphically rich and interactive HTML5 websites to come from developers:
Note this video uses the HTML5 video tag (with the H.264 codec) if your browser supports it, and falls back to other methods otherwise. It’s a good example of same markup in action.
Performance through the power of the whole PC
Today, people expect less from a website than they do from native applications in terms of power, richness, responsiveness, and interactivity. With the third platform preview, we continue to deliver on the promise of a fully hardware accelerated browser where all of the support for text, graphics, and media uses the underlying hardware through Windows, making the full power of the PC available for the Web. Using the power of the whole PC shatters the previous constraints that limited websites.
JavaScript is one component of browser performance, and Webkit Sunspider is one measure of script performance. The latest platform preview shows how IE9’s JavaScript engine continues to get faster. Here’s the chart:
You see the progress with this zoomed-in chart, showing just the IE9 platform previews and the pre-release versions of other browsers:
Looking at the differences between the script engines’ performance, you see the performance gaps between the fastest script engines are now less than 50 milliseconds – and that’s executing several million script instructions during the benchmark test. This difference is already under any human perception threshold, and we’re not done yet.
Many sites spend lots of time in subsystems other than JavaScript. If browser performance were entirely attributable to JavaScript, the performance on the test drive site samples would look like the Webkit Sunspider graph; that’s not the case. You can expect that we will continue to focus on real-world performance and not optimize for any specific benchmark.
Introducing hardware accelerated canvas, video and audio
With the third platform preview, we introduce support for the HTML5 Canvas element. As you know our approach for standards support is informed both by developer feedback and real word usage patterns today, along with where we see the web heading. Many web developers have asked us to support this part of HTML5 and we definitely took this feedback into account as we prioritized our work.
Like all of the graphics in IE9, canvas is hardware accelerated through Windows and the GPU. Hardware accelerated canvas support in IE9 illustrates the power of native HTML5 in a browser. We’ve rebuilt the browser to use the power of your whole PC to browse the web. These extensive changes to IE9 mean websites can now take advantage of all the hardware innovation in the PC industry.
Try the Asteroid Belt sample and Fish Tank sample on the test drive site to see hardware accelerated Canvas in action. Together with Amazon, we built a book shelf sample showing the potential for bringing the richness of hardware accelerated canvas into an interactive e-commerce experience.
Similarly, we partnered with the Internet Movie Database (IMDB) to build the Video Panorama sample to demonstrate the possibilities for bringing hardware accelerated HTML5 Video and graphics interactivity together to create new applications and experiences. Our focus is on delivering a complete, highly interoperable implementation of canvas, video, and audio that makes the most of the full power of the PC.
To help you better understand how these samples work, we created videos that provide a look “under the hood” for Fish Tank, Amazon Shelf, and Video Panorama. As the browser uses more of the hardware, your experience depends on the hardware you have, just as it always has. With hardware accelerated graphics, the graphics card and driver combination play a significant role in how you experience the various examples and benchmarks.
The PC hardware ecosystem has made huge advances over the last few years. Today’s GPUs provide up to 10 times the computing power of CPUs, a clear trend in GPU processing power compared to CPU over recent years. Given how important the web is, our focus is making that power available to web developers. With IE9 developers can run the same, W3C standards-based markup as in other browsers – just faster, taking advantage of the underlying hardware. The result of using the power of the whole PC to browse is a new class of web application without many of the limits on today’s sites.
The first two platform previews demonstrated hardware acceleration of text, images, and vector graphics. Preview 3 completes the media landscape for modern websites with hardware accelerated video, audio, and canvas. Developers now have a comprehensive platform to build hardware accelerated HTML5 applications. This is the first browser that uses hardware acceleration for everything on the web page, on by default, available today for developers to start using for their modern site development.
Same Markup
As we have said before, web browsers should render the same markup – the same HTML, same CSS, and same script – in the same way. That’s simply not the case today across many browsers and many elements of markup. Enabling the same markup to work the same across different browsers is as crucial for HTML5’s success as performance. Our investments in standards and interoperability are all about enabling the same markup to just work. When developers spend less time re-writing their sites to work across browsers they have more time to create amazing experiences on the web.
The third platform preview continues to support more of DOM and CSS3 standards that developers want. Examples here include DOM Traversal, full DOM L2 and L3 events, getComputedStyle from DOM Style, CSS3 Values and Units, and CSS3 multiple backgrounds. And as with each previous Platform Preview, the Developer Guide includes a list of all available features. We continue to work with the standards bodies and the browser web community to deliver same markup.
Same Markup: Video, Audio, and WOFF
At MIX we showed the potential for hardware accelerated video on the web. On the test drive site you can try out several examples of IE9’s support for HTML5 video and audio tag support. You can see for yourself how sites like YouTube for HTML5 works with video playing through the GPU. Here’s an example:
Note that while the video element in the HTML5 standard does not detail support for specific codecs, developers can still use the same markup to achieve the results they want. Coding practices should focus on testing for capabilities, not browsers, and providing the right fall backs for older browsers.
Also included in the third platform preview is support for using the Web Open Font Format (WOFF) through CSS3 font face. We were excited to work with Mozilla and Opera to submit the WOFF file format to the W3C, and in IE9 to bring high quality font support to the web in a way that is friendly to web designers, font foundries, and end users. As an industry we still have work to do for same markup with same results for font support.
Like all of the text rendered in IE9, the support for WOFF makes the most of the underlying hardware and Windows DirectWrite for high quality rendering with sub-pixel precision, resulting in smooth, crisp text across font sizes and browser zoom levels. In a recent article, Richard Fink from A List Apart wrote how, “The font rendering in IE9 is worlds apart from what we’ve all come to expect.”
Of course, the importance of WOFF support is having the same markup provide the same results for text and typography - results developers and designers can depend on. Here’s an example from the test drive site that shows a selection of WOFF fonts in IE9 and in latest shipping versions Firefox and Chrome. Some of the differences you’ll see if you try this example yourself are more precise layout of text and sharper characters at large font sizes and when zoomed.
Same Markup: JavaScript and ES5
Same markup includes running the same JavaScript code with the same results. The Chakra JavaScript engine in IE9 significantly improves support for ECMA-262: ECMAScript Language Specification standard, including features new to the recently finalized Fifth Edition of ECMA-262 (often called ES5 for short). This support for ES5 includes new array and object methods, as well as other language enhancements for working with strings and dates. The test drive site includes samples where you can try new array methods and play a game built with new ES5 capabilities. You can learn more about how we used ES5 arrays with a video look “under the hood” for the Tile Switch game.
Following through on having same markup across the web requires comprehensive and consistent tests. Unlike the standards body behind CSS and HTML, the JavaScript standards body has never had a place where the community could contribute tests. We’re working as part of the “TC-39” community with Ecma to create an official test suite for ECMAScript, sponsored by Ecma. In anticipation of this site, we’re making tests available on the IE Testing Center now for feedback via Connect.
Same markup: IE Testing Center and other tests
Some people use Acid3 as a shorthand for standards. Acid3 tests about 100 fragments of a dozen different technologies. Some are still in “under construction.” Some of the patterns, like SMIL animations, are inconsistent with others, like CSS3 animations, and need to be reconciled. As we continue to implement the standards that developers have told us they value most, the Acid3 score continues to rise, and we’re not done. Here’s a screenshot of how today’s IE9 Platform Preview runs today’s Acid3 test, going from 68 in the last platform preview to 83:
With today’s update to the platform preview, we have also updated the IE Testing Center, adding another 118 test cases which we are contributing to the appropriate web standards working groups at the W3C. In addition, we have written 1309 JavaScript test cases and are making those all available to the web development community. Another blog post provides details about them.
Performance measurement in web pages
Enabling developers to accurately measure website performance is important to delivering great HTML5 applications. Today, developers can measure different aspects of performance on their own machines with the Developer Tools; they can’t, however, measure the performance their users actually experience. Today, many sites develop their own libraries that try to measure live performance on web pages. The problem is that these libraries actually slow down the pages for consumers and measure inaccurately, driving the wrong behavior for developers.
We believe that the WebTiming specification is a good conceptual foundation for solving this problem. We’re in conversations with the W3C HTML5 standards body and folks at Google and Mozilla about this, and how we can all work together to make WebTiming happen in an interoperable and standardized way, sooner rather than later. We will work closely with the W3C and its members over the coming months to get this into an official working group and build consensus for a proposed specification while continuing to work together to ensure that the same markup works across browsers.
In order to keep making progress in the interim, we’ve included early support for these ideas in the IE9 preview. This is early work for sure, and following convention, IE uses a vendor prefix (-ms) on the namespace because the spec is genuinely under active construction. You can take a closer look at how this works in the WebTiming sample on the test drive site. We’ll talk more about this topic in a future blog post.
Test Drive IE9 Platform Preview Today
We appreciate the continued feedback about what we need to fix (at Connect, on the HTML5 testing listserv) as well as the feedback about what we’ve done right.
Our continued ask, is that you download the latest preview, try the samples on the test drive site, and try your own sites. Send IE9 the same markup that you give to other browsers. The IE7 compatibility view built into IE9, which some sites may run in, does not offer the best performance possible. If you still have sites that run in IE7 compatibility mode we recommend that you move those to IE9 standards mode. We want sites to get the full performance benefits of IE9 that come with running in IE9 standards mode. We also want your feedback from handing IE9 the same markup you hand other browsers.
The platform preview installs side by side with Internet Explorer 8 so that you can try it without replacing the full version of Internet Explorer that comes with Windows. This third release of the Internet Explorer 9 Platform Preview will install over the prior versions. There is no need to uninstall the earlier builds before installing the third. You’ll also find more information on what’s included in this release of the Platform Preview in the Release Notes, including known and resolved issues.
Thanks
Dean Hachamovitch
General Manager, Internet Explorer
Comments
Anonymous
January 01, 2003
The comment has been removedAnonymous
January 01, 2003
The comment has been removedAnonymous
June 23, 2010
I love to see this progress, I'm very excited for Microsoft. However I'm really puzzled by the fact that people have to register on MSDN in order to see this news. Surely blogs should be public, otherwise you're not really informing the general community are you?Anonymous
June 23, 2010
Any plans to allow any other font types into IE9? Because, honestly, as all other browsers support TTF and IE currently needs EOT, it's not really an improvement to let all other browsers still support TTF while IE9 now supports WOFF. It does absolutely NOTHING for developers unless Microsoft can cajole Opera, Google, and Apple to also support this format. You've merely replaced one "gotta do this tweak for IE" with two: "gotta do this for IE9+, and this for IE8-". I'm looking forward to test-driving your HTML5 capabilities, though. Hopefully it can pass my CSS/SVG/XML test page I've written - earlier versions didn't yet! (velociraptorsystems.com/.../glass)Anonymous
June 23, 2010
I second StonyUK's sentiment...glad to see the strides IE is making.Anonymous
June 23, 2010
@StonyUK who has to register to see what now? I'm browsing this blog without signing into anything, and i was able to download the preview without logging into anything either.Anonymous
June 23, 2010
developers don't be so happy yet with this announcement. untill it ships they will make sure to have at least 49 bugs :(Anonymous
June 23, 2010
@Philip Kahn: Opera and WebKit-based browsers will add support for this format.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Mr. Potato is an egg?!Anonymous
June 23, 2010
@Victor didn't they say WebM will be supported by a downloadable codec? or was that VP8 (same diff?)Anonymous
June 23, 2010
@Viktor never assume: if you have a WebM VP8 codec it will play from the HTML 5 video tag just like H264 - the IE team announced this last month.Anonymous
June 23, 2010
Great! I'm really happy MS back to the game again! But... IE9 should be a mandatory update for all XP/Vista/W7 users even without asking about permission! IE6 / IE7 /IE8 - let it go gracefullAnonymous
June 23, 2010
@Philip Kahn: Firefox already supports WOFF. Other browser vendors are working on their implementations. Font vendors are also supportive of the format.Anonymous
June 23, 2010
Google Chrome 6 will support the FileReader and FileWriter APIs to read and write local files in webapps. Firefox already has FileReader since version 3.5. When will IE have these? These two APIs are must haves for webapp developers.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Oh, so you have OneNote 2010 installed?Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
@Mark: You can already upload local files, so what's the difference, security-wise, for FileReader? And FileWriter uses a folder specific to your webapp. Google Chrome will have these shipped by August, and Google is very serious about browser security (even more than Mozilla), as I'm sure you're aware. Mark, next time read the specs before you post here these kind of comments.Anonymous
June 23, 2010
Is this a joke? When will you give up??Anonymous
June 23, 2010
Mark: You can already upload local files, so what's the difference, security-wise, for FileReader? And FileWriter uses a folder specific to your webapp. Google Chrome will have these shipped by August, and Google is very serious about browser security (even more than Mozilla), as I'm sure you're aware. Mark, next time read the specs before you post here these kind of comments.Anonymous
June 23, 2010
@GregT: right on.Anonymous
June 23, 2010
Why does the title bar still say Platform Preview 1??????Anonymous
June 23, 2010
In IE9, you should put a feature that allows for automatic refresh every so often. I know that that would help me!Anonymous
June 23, 2010
I'm unable to install the preview. It gets to the point where it says installing prerequisites - it gets to KB2120976 and stalls out saying "The installer has encountered an unexpected error installing this package. this may indicates a problem with this package the error code is 2739.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Andy, you're clearly new at this security thing. You should probably do some more study before you try to participate.Anonymous
June 23, 2010
Jon: I don't hate IE. I just need the FileReader and FileWriter APIs for my next webapp (to be shipped by September). If IE won't support these two APIs, I'll tell my users to install Google Chrome Frame. Simple as that.Anonymous
June 23, 2010
The installation of the preview is getting stuck when downloading a prerequisite update KB2117917.Anonymous
June 23, 2010
Andy, you're clearly new at this web thing. You should probably learn why telling users to install another browser is a good way to flag to them that they should find a better site.Anonymous
June 23, 2010
I love seeing the work IE Team is doing to stay competitive; each developer preview keeps getting "much better," not just "a little better." The hardware acceleration, full stride with HTML5, fonts, and scripting performance is flat out awesome. I have, and many others, steered clear away from IE after seeing the lack of performance and poor rendering (mixed proprietary, non-standard conformity) over each release, but now I'm anxious to see the final product. I'm crossing my fingers for your teams success. =)Anonymous
June 23, 2010
Mark: "Andy, you're clearly new at this security thing." :D Too funny! I've developed my fist website in 1993, when did you? BTW, that was a rhetorical question – this is my last comment on this post. Don't bother to reply.Anonymous
June 23, 2010
Mark: Google Chrome Frame is a plugin for Internet Explorer. NOT another browser. You are only showing your ignorance. Please stop.Anonymous
June 23, 2010
@Kevin Scott Check this out: stackoverflow.com/.../refresh-http-header Feel free to add it to any page you want to refresh constantly. It's worked since....the dawn of time? :PAnonymous
June 23, 2010
What about WebGL???Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Andy, ChromeFrame is a browser by any technical definition of the term (rendering stack + network stack + script engine)Anonymous
June 23, 2010
@Not Kevin Scott I think he's reffering to a user-set automatic refresh like in OperaAnonymous
June 23, 2010
@mitch-- that site also says that MP3, AAC, MPEG4, and H264 aren't supported either. The "test" site is bogus, and its scoring is arbitrary.Anonymous
June 23, 2010
@Paul No, ChromeFrame is not a browser, is basically WebKit wrapped arround a Active Document Server (msdn.microsoft.com/en-us/library/8ha0eexy%28VS.71%29.aspx) (i.e. the same thing Trident/Mshtml.dll is) ChromeFrame USES IE network stack! However, ChromeFrame does not replace the needs to support HTML5 in trident.Anonymous
June 23, 2010
@hand, @Sylvain [MSFT] : Ah, didn't notice any official announcements on their parts about WOFF. In particular if Webkit doesn't support it, it really would mean WOFF was a useless change from EOT (not really changing the status quo). I did know Gecko supported it, though. Is it better to report specific site crashes casually here, or file formal Connect reports? It seems on preliminary analysis that pages delivered as XML (application/XHTML+XML) crashes this PP.Anonymous
June 23, 2010
@João Serra: While it may still be true that ChromeFrame uses the WinINET stack (they were still debating last time I looked) they have their own rendering engine, script engine, content parsers, image decoders, etc. From a security point of view, it has virtually all of the attack surface as a browser (actually more, due to the disconnect between the security model of the addon and its host). No intelligent and security conscious network admin will let ChromeFrame anywhere near their network.Anonymous
June 23, 2010
"No intelligent and security conscious network admin will let ChromeFrame anywhere near their network." Most network admins only permit IE6 on enterprise networks, so if they trust IE6 and its insecurities and not ChromeFrame, then they are incompetent...Anonymous
June 23, 2010
Congratulations, guys! I love the enormous progress you're making on the browser engine. However, you should also modernize the UI of IE and bring it on par with other browsers. Chrome, Opera, and Firefox 4 all will have beautiful UIs with smooth lines and animations. For example, IE8 now has the ugliest tabs of all modern browsers. In contrast - try reordering tabs in Chrome - it's just beautiful. And please remove the deep sunk border of the IE frame by default (set DOCHOSTUIFLAG_NO3DBORDER) - it looks so heavy and 1995...Anonymous
June 23, 2010
Why we should bother if visitors have WebM codec? They won't have it (most of them). Why not bundle this thing? I can't see the point.Anonymous
June 23, 2010
Roland: If you'd like, you can remove the 3D border: www.enhanceie.com/.../no3dborder.regAnonymous
June 23, 2010
In a previous post [1] I asked about the differences in format support between the AUDIO and BGSOUND elements, to which you responded, "The AUDIO tag will support AAC and MP3. BGSOUND will continue to support WAV." Is that still Microsoft's position, and exactly why are wave files not supported by the AUDIO element, given that it is already supported by your browser? [1] blogs.msdn.com/.../another-follow-up-on-html5-video-in-ie9.aspxAnonymous
June 23, 2010
Paul: Internet Explorer has more detected security holes than Google Chrome: secunia.com/.../21625 An intelligent and conscious network admin won't allow Internet Explorer to be installed. He/she will install Google Chrome instead. Yeah, the browser one. There. I said it.Anonymous
June 23, 2010
@Roland I can't stand the 3D Border in IE. It's always one of the first things I remove in the registry when I use IE.Anonymous
June 23, 2010
@Dan: The test is not bogus, in fact it is open source and you can look at the code. It is testing the canPlayType() method of the video element and that is apparently not correctly returning a true value. Thus the feature is not detected in JavaScript.Anonymous
June 23, 2010
I raise my hat to another giant leap for the upcoming IE9. Unfortunately, the .getElementsByName bug - a huge inter-browser compatibility issue - remains. After posting on comment on IEBlog a few months ago, someone hinted that posting a bug on microsoft's connect bug reporting tool would help. connect.microsoft.com/.../getelementsbyname-should-return-all-html-elements-with-a-name-tag This was done on May 7. It has however apparently received very little attention from the IE team (other than the generic "thanks for your reporting" reply). I'm not sure what else I can do.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
I'd prefer for MS to just build IE 9 from Webkit and contribute to Webkit to provide additional functionality.Anonymous
June 23, 2010
I'd prefer for MS to just build IE 9 from Webkit and contribute to Webkit to provide additional functionality.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
@Steve Like they said in the post, browsers like Webkit have to provide lowest common denominator functionality since they're cross-platform. Also, which Webkit should they use? There isn't exactly one Webkit out there. They all have different degrees of standards support. It's not like Chrome and Safari are equivalent because they use Webkit.Anonymous
June 23, 2010
@raffi: "browsers like Webkit have to provide lowest common denominator functionality since they're cross-platform..." "They all have different degrees of standards support. It's not like Chrome and Safari are equivalent because they use Webkit." :D Wow! So much FUD in just one comment! Maybe someone has the patience to reply to this guy?Anonymous
June 23, 2010
@raffi: Here's THE WebKit – the ONE both Google Chrome and Safari are based on: http://nightly.webkit.org/Anonymous
June 23, 2010
Why can't I maximize the HTML5 video in this blog post in Google Chrome 5?Anonymous
June 23, 2010
Whats FUD? There are a ton of forks and ports of Webkit floating around, especially in the mobile space. So when someone says "just use Webkit." Which one? And the lowest common denominator comment is totally true. For example, if a web browser wants to support hardware acceleration they could use OpenGL which is cross-platform. But then how are they going to support printing or handle font rendering? Firefox will use Direct2D like IE9, but Direct2D isn't available outside of Windows. So they will have to make some tough engineering decisions, and may have to maintain both D2D and OGL renderers, plus whatever they use for printing. So either you go the lowest common denominator route, or you take on a lot of extra work. Neither of which IE has to bother with, since its only on Windows.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
@Phillip Kahn - if you read the release notes, IE9 PP3 supports WOFF, EOT, TrueType, and OpenType. That covers more than enough.Anonymous
June 23, 2010
I'm also getting an error code 2739. While trying to download. I tried to redownload it about 5 times, so I don't think it's a download problem. Any suggestions?Anonymous
June 23, 2010
raffi: You yourself said:
- "Firefox will use Direct2D like IE9 [on Windows]" Both WebKit and Firefox are implementing features using OS-specific APIs when necessary for performance reasons, so your lowest-common-denominator comment is misleading and irrelevant. And again, here's ONE WebKit for the desktop – the ONE both Google Chrome and Safari are based on: http://nightly.webkit.org/ Steve was talking about Internet Explorer on the desktop, wasn't he?!
Anonymous
June 23, 2010
@Gaurav: The HTML5Test.com wraps canPlayType and requires a response of "probably" unless the client is an IPhone, in which case they also accept "maybe". The IE9 PPB returns "maybe", but because of hardcoded browser detection requiring IPhone only, this is incorrectly classified as a failure by the test site. You'll need to talk to the creator of this "standards test" to determine why they believe that returning failures based on the sniffed browser is a reasonable thing to do. @Mathieu: If I recall correctly, for your GetElementsByName issue, you're hitting this only on elements that are not defined to have a NAME attribute in the HTML spec. Is that correct? Is there a reason that using an attribute selector wouldn't work?Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Great work on IE9! I'm really looking forward to being able to rely on this being the baseline. I'll be able to make amazing apps. Only things I'm missing are workers (all that javascript performance is no good if it blocks the UI), and webgl (although admittedly it needs to bake a little more before it is ready). Maybe you guys could get on board with the khronos group and help finalize the webgl spec?Anonymous
June 23, 2010
installer stuck at downloading KB2117917Anonymous
June 23, 2010
For those running into the 2739 error, please try installing the updates manually from the installation help page and launch setup again. ie.microsoft.com/.../Default.html If you continue to have installation issues, please file a bug in Connect. Log file information is very helpful which you can get by running setup with the /l* [logfile] parameter.Anonymous
June 23, 2010
@Jesse Mohrland [MSFT] thanks, that workedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Great Job IE Team! It's great to see the SunSpider test results coming close to 100 and the Javascript engine performance get close to the pack leaders in this space. It's also great to see IE leading and creating W3C test suits so other browser makers can conform. Html 5 video is finally here in IE 9 as well! Lots of really good stuff and the hard work the team is doing is really showing. I agree with Jason (above) about the more frequent releases and the fact that the other borwsers are not "preview releases" but rather production releases. What really happens is that the user see new features being implemented and websites working better (due to html 5 support) in other browsers and not at all in IE. So seriously think about beta release of IE 9 and frequent updates. Enough of this "preview" thing. You guys need to change your focus to html 5 features that other browsers are focused on because those are the features that will eventually count. Features that are supported by more than one browser will start to be implemented by websites and browsers that don't work will end up getting a bad name. Check out Google's Html 5 playground http://www.html5rocks.com/ This site simply doesn't not function with IE 9. Besides, the features being shown off there are the features that "matter". Incredible GPU acceleration is great but unless other browsers come close those aspects will not be used by anyone. Frankly I've had enough of the 60fps versus 12fps thing. I mean I get it and I love it but let's move beyond fps for now and do what others browsers are doing too. When are we going to see html video fullscreen support?Anonymous
June 23, 2010
Hello! Greetings from Finland. ;) Problems with installation? If the problem is KB2117917, download the "Platform Update Supplement Beta" from here: support.microsoft.comAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
At first this looks great but I can't take this blog post seriously. If someone are showing demos of html 5 stuff in several browsers and it only works fine in one (in this case IE) it can't be standard code. Did take a quick look at the code for the fish tank example and it look quite good tough... but something is fishy here :) But as mentioned above, me also is tired that MS is always going there own way on what to support or not. When working with front-end development on a daily basis all I want is a standard code base. Good work with the development of IE9, will be nice to see it ship in couple of years when everything is changed on the web and we half to wait for IE11 to support the new technics, and the spiral begins all over again. Sorry for bad english, hope the message is getting trough :) Happy coding!Anonymous
June 23, 2010
Same Markup "As we have said before, web browsers should render the same markup – the same HTML, same CSS, and same script – in the same way. That’s simply not the case today across many browsers and many elements of markup." Did I miss something? Even CSS2 support is incomplete in IE8. We don't want fancy graphics that only work in one browser, which only works on a small percentage of the computers in use. We want a browser that actually renders standard markup and CSS correctly, like so many other browser vendors produce.Anonymous
June 23, 2010
@Dan: since this site can detect Chrome/Chromium, Opera and Safari properly, I'd say that it's simply that one has to use a different markup in IE 9 to detect the browser's capabilities. Which goes against the 'same markup' promise. And, since IE 9 is the odd one out, I'd say it's a problem with the browser - not the site.Anonymous
June 23, 2010
@tweaksource Actually CSS2 was already feature complete in IE8 (allthough not flawless)Anonymous
June 23, 2010
Sadly this version crashes for me. The first two previews did not. Don't know if this helps, but the problem details show the following: Problem signature: Problem Event Name: APPCRASH Application Name: iepreview.exe Application Version: 9.0.7874.6000 Application Timestamp: 4c1c6efb Fault Module Name: msvcrt.dll Fault Module Version: 7.0.7600.16385 Fault Module Timestamp: 4a5bda6f Exception Code: c0000005 Exception Offset: 00009c7f OS Version: 6.1.7600.2.0.0.256.4 Locale ID: 2057 Additional Information 1: 0a9e Additional Information 2: 0a9e372d3b4ad19135b953a78882e789 Additional Information 3: 0a9e Additional Information 4: 0a9e372d3b4ad19135b953a78882e789Anonymous
June 23, 2010
I solved the issue with the installation of the preview getting stuck on the downloading of prerequisite KB2117917 bij downloading the prerequisite straight from the Microsoft website. After the required reboot of that prerequisite the installation functions correctly.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Great news to see that Canvas is being implemented - I'm quite excited by IE9 now! Keep up the good work, and thanks. One plea: please, please implement WebGL. As someone above said, Canvas really isn't up to scratch when it comes to 3D (even with HW accel), and webgl looks pretty superb. Even some kind of wrapper that allows people to write webgl markup and have it rendered via DirectX or whatever it is would be great.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
I love reading your blog posts about IE9 and I really look forward to the final version! Great work so far!Anonymous
June 23, 2010
OK, so when color profile (ICC) support is coming to IE? With implicit sRGB for non-tagged images?Anonymous
June 23, 2010
Canvas support is much appreciated.Anonymous
June 23, 2010
@"Mitch 74": « don't forget about WebGL support: canvas ain't so good at 3D. » Canvas and SVG are pretty good at creating 3D stuff, it just might not run very well in real-time. « replace PDF, an industry and ISO standard, with XPS, a patent-encumbered proprietary format? » The container structure of XPS is an ISO standard and the format itself is standarised by ECMA. Also, Adobe has likewise a number of patents covering technology in their format, just read the legal notices in the Adobe Reader about window. @raffi: « Firefox will use Direct2D like IE9, but Direct2D isn't available outside of Windows. So they will have to make some tough engineering decisions, and may have to maintain both D2D and OGL renderers, plus whatever they use for printing. » AFAIK, they will probably maintain implement even more renderers like OpenGL ES and DirectX 9, though most of this won't be available for common use anytime soon. HW acceleration is not going to be activated per default for Firefox unless they find some way to detect quirky hardware/drivers and such (there are plans for creating a community-supported project for creating a DB that might help in this).Anonymous
June 23, 2010
On the subject of Windows Installer error 2739, please see Heath Stewart's blog about errors 2738 and 2739: blogs.msdn.com/.../windows-installer-errors-2738-and-2739-with-script-custom-actions.aspx They stem from an incorrectly-registered VBScript (2738) or JScript (2739) engine. I opened RegEdit, deleted HKCUSoftwareClassesCLSID{F414C260-6AC0-11CF-B6D1-00AA00BBBB58} and the install worked the second time around. Rob Mensching, the author of the Windows Installer XML (WiX) toolkit, recommends avoiding VBScript and JScript custom actions: blogs.msdn.com/.../136530.aspxAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Please please please implement HTML5 forms. You will make the world a better place if you do that!Anonymous
June 23, 2010
Congratulations Microsoft! If you support workers and web sockets, I will throw away other browsers ;) CheersAnonymous
June 23, 2010
@Indrek Actually the maximum score of acid3 is a 101, their is a favicon in combination with a 404 Not Found-error check which they couldn't find a way to test for. But I think all modern browsers have that down, including IE8 so it seems from first glance.Anonymous
June 23, 2010
Anyone else having issues with the new preview not rendering pages after a couple of second? The pages are loaded, you can see the title and hyperlinks change as you browse around. But the actual pages aren't being displayed, 9 times outta 10 you're left staring at the home page.Anonymous
June 23, 2010
@ieteam: I'm trying to install this Platform Preview but I get error 2379, how can I fix it? :) thank you!Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Fantastic work! It seems like IE will be after long time again best browser...Anonymous
June 23, 2010
i always use ie!!!! great!!Anonymous
June 23, 2010
@Paul Smith I agree I'm having the same problemAnonymous
June 23, 2010
Thanks for supporting canvas! Maybe soon I can retire my fallback to moving divs, for my HTML5 game.Anonymous
June 23, 2010
Another vote for HTML5 Forms support. It can really make a difference in webapps and sites.Anonymous
June 23, 2010
@GregT Don't worry about such things. Their won't be an IE9 for Windows XP, which was confirmed by Microsoft. So old versions of IE will take a very, very long time to die.Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
blog-imgs-42-origin.fc2.com/.../dora_css3.html once transform(rotate) is implemented, doraemon's legs and hands should look better :PAnonymous
June 23, 2010
Love it! However, I ddid find a bug in the <video> implementation where moving a node causes it to fail. If you attempt something like: document.getElementById('somediv').appendChild(document.getElementById('somevideo')); It dies. Hope that helps!Anonymous
June 23, 2010
Canvas, audio, and video support are huge, kudos to the IE9 team. However, what I really need now is Web Workers. Without them it's hard to do anything useful with the insane computational speed of all these new JS engines. So, what are your plans in regards to Web Workers?Anonymous
June 23, 2010
great! work guys keep it up!!!Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
Hello, Kudo for the improvement that Microsoft is doing in the standards. I have only one thing to say. Be careful... HTML 5, Javascript, canvas is nice, BUT not everybody has a powerful configuration, HW accelaration support, etc... I have just checked the Asteroid Belt Sample on my Firefox 3.6.3 on Linux (Fedora 13 64bits) with no HW support for my graphical card, well the minimum HW support. And it blocked for several minutes my workstation. Until Firefox asked me to stop a javascript which was using all the resources of my workstation... I was doing way less than 1 FPS, it was like 1FPM !!! or even less! And only by this experience, I don't want to test the other things you presented on this page... So, it is good in one way, but in the other way, the web is a place where everybody must experience something similar, and not have some discrimination on the OS/HW/Browser a user is using. And if the trend is going that way... I am sorry, but does not help the users at the end... it will only be a show for techies in lack of sensation. Cheers, AlessandroAnonymous
June 23, 2010
I second Ronald's thoughts on HTML5 Canvas support. Good work!Anonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
The comment has been removedAnonymous
June 23, 2010
@Bronek: ICC color support (v2, v4) was introduced in the first Platform Preview of IE9.Anonymous
June 24, 2010
@ErikLaw For all browsers I only accept "probably" as indication that the browser supports a codec. The only exception is the iPhone where I accept "maybe" as indication for codec support. If I remove that exception for the iPhone, IE would still have to report "probably" to score any points, just like the other browsers do. I can't use the same workaround for IE. If I do, I would also have to report MPEG4 support for IE, which is a false positive. At the moment you simply can't distinguesh between MPEG4 and H.264 support on IE9, which needs to be fixed. Cheers, Niels Leenheer html5test.comAnonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
NEils: so you admit that the html5test site is a lie then? Good.can you post that up on the site itself? kthx!Anonymous
June 24, 2010
this is awesome :)Anonymous
June 24, 2010
@Walker: That site tests a lot of feature that are not part of HTML5 but counts them as if they were HTML5 features. Do you not think this is what people refer to as a "lie"?Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
Dave: Useless rant without anything worth time. And stop misrepresenting yourself as representative of "world". Thanks, Real WorldAnonymous
June 24, 2010
Hi IE Team, This is great work in proactively implementing HTML5. A lot of developer friends and I were wondering if HTML 5 wouldn't realistically be adopted for another 10 years because of past IE adoption track records. This looks like it'll happen a lot sooner and that's extremely exciting. However, PLEASE PLEASE PLEASE PLEASE DON'T RELEASE IE9 UNTIL IT PASSES (passes mean 100 btw) THE ACID3 TEST!!!!! Sorry for caps, it's annoying, I know. Also, please refer to quirksmode to try and implement as much of the other browser features as possible. Over the past few releases of IE, my websites have "broken less" with each release. However, all of the other browsers don't break. So, if you do pass (100) the acid3 test and implement features in the html5 standard, couldn't you "recall" all of the browsers that "are still reeking havoc" in the internet. Namely, IE6 (still a-grade). But I wouldn't complain if ie7 and ie8 died with it. This is a promising advance and kudos on being proactive with html5, but we could really just use a cleanup of past mistakes at the moment. Thanks, SteveAnonymous
June 24, 2010
@Alessandro: Firefox 3.6.3 (on windows) will run the Asteroid Belt sample, although you have to constantly move your mouse to actually see the results. And firefox having no HW support on linux is a problem you need to bring up with them not blame IE 9 for supporting available hardware just because the platform/OS/browser you chose doesn't support it. Opera, Safari, IE9 all run the sample just fine. Firefox is buggy, and chrome is horrid, but it works.Anonymous
June 24, 2010
@IESucks: Webkit has it's own share of bugs and problems, and if you report bugs to them, they don't get fixed. The webkit team is the absolute worst.Anonymous
June 24, 2010
I try 'sudo apt-get install silverlight' but no video show. Is microsoft broken?Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
I am really excited to see all these new features in IE9! It will be really good that IE stops being the Ugly Duckling of the browsers. As someone commented above it would be nice that besides all the hard work on GPU features you start focussing now on HTML5 so you are not the only future browser with bad numbers here whencaniuse.com Believe me when I say that the IE team always had good intentions for standards support for IE7 (http://bit.ly/dn1NKA) and IE8 (http://bit.ly/aG3ezW) but the implementations weren't as good as their intentions in the end. I hope this time is different for IE9. One thing I recommend the IE team though is to stop using misleading demos. Comparing your GPU speed of your future browser against current browsers that use only CPU is more of a marketing strategy than a sound technical comparison. Please compare your demos against other GPU implementations... or just stick to the numbers or the acid3 test improvements and developers will love your effort without thinking there's something smelly behind it.Anonymous
June 24, 2010
Way to step your game up IE team. Way up.Anonymous
June 24, 2010
Just saw this post about the ES5 support perfectionkills.com/jscript-and-dom-changes-in-ie9-preview-3 and your tables at the testing center in more detail. Hats off for this. Congratulations!Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
@RobertM: What kind of bug reports? Ones with reduced test cases or the ones similar to "when I clikz buttan it duz not workz". @D.S: In the newest revisions of the site, the supported codecs just give separately counted "bonus points".Anonymous
June 24, 2010
Congrats to the guys and gals at Microsoft. Good to see the company taking the browser much more seriously. If the jump between the previous IE and IE8 is any indication of things to come, I can't wait to see IE9 unleashed!Anonymous
June 24, 2010
@D.S: >> How do you know that none are planned for IE9? "So, of the 23 modules in SVG, we are implementing 20. The three that we are not shipping within IE9 are SVG Fonts, Filters, and SMIL." from www.w3.org/.../interview_ieAnonymous
June 24, 2010
@EricLaw[MSFT] Thanks for clearing up the issue with html5test.com. I have reported it as a bug. See github.com/.../issuesAnonymous
June 24, 2010
Now, You need a killer debugging tool for IE9. Developers go to a browser with better debugger. IE dev toolbar has some catching up to do with Firebug and Chrome Dev toolbar.Anonymous
June 24, 2010
And this MSDN blog requires a serious commenting system. I am sorry to say this but, what you have here is rubbish at best.Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
Thanks for the great news.Anonymous
June 24, 2010
@MarcSil [MSFT] "html5rocks.com checks for IE using navigator.userAgent.match(/MSIE (d+(?:.d*)?)b/)" I think that instead we are enabling chrome frame for most of the sections because it was impossible to run it well with any of the IE versions. I will be more than happy to remove any filter so you can run it with IE9. "We are reaching out to Google and other site developers to request that their samples work across browsers." I didn't receive any email yet but please do. I want IE to be present along with all the other browsers.Anonymous
June 24, 2010
@NielsLeenheer Your test is bogus because it does not detect supported codecs which it claims to do. You give bonus point for returning "probably" on a canPlayType call and not for actual support of the codecs themselves which your test suggest it does.Anonymous
June 24, 2010
@José I tracked this down to a possible issue with video drivers on certain laptops with ATI or Intel graphics cards. Try uninstalling or updating the driver to see if the problem can be resolved.Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
@Steve. JS processing doesn't happen on the GPU, it happens right there on the CPU. What happens on the GPU is the rendering of text and graphics. All processing still happens on the CPU.Anonymous
June 24, 2010
@Jesse Mohrland [MSFT] I ran into the same problem as José did. This is on Windows 7 x86 on a notebook with indeed an associated ATI graphic chip. There's no driver update available for that one (no surprise) and uinstalling the driver isn't an option of course. But the problem as such vanishes as soon as you uninstall the "helper" update KB2028560. Strange. FWIW, FreudiAnonymous
June 24, 2010
<<<IE6,7,8 all live quite happily right now on XP (and have done for a decade) with no end in sight in the next decade>>> Wrong. XP SP2 is completely out of support on 7/13/2010. SP3 is in extended support (security fixes only) until 4/2014, which means that it gets security fixes only, not functionality fixes. Windows 7 has the fastest uptake of any operating system in history, and already is the most popular OS on Valve's Steam.Anonymous
June 24, 2010
@Mitch74 "IE 9 has h.264 built-in, we're told;" I can't remember we are told that nor do I know if that is the case. "If IE doesn't, it doesn't follow the expected syntax. So, the test is right, and there's a bug in IE9. Before you claim a test to be bogus: read the spec." If you were right about the syntax that would still only mean that IE9 does not comply with the canPlayType syntax. So the use of canPlayType is what is given bonus point for. Not for actual support of the codec as the test page claims.Anonymous
June 24, 2010
@hAl: Yes, it was announced at MiX 2010 (and on the blog) that IE9's Video tag will natively support h264 in an MPEG4 container. It will also support WebM if the user has installed that codec separately. For the Audio tag, MP3 and AAC are natively supported.Anonymous
June 24, 2010
Gee...you demo IE9 on WindowsXP & I'm watching on Vista Home Premium build 6002? XP is so 8-years ago...I'm so proud of Vista that it may take a while to go Window7!Anonymous
June 24, 2010
Backyard, IE9 does not run on Windows XP.Anonymous
June 24, 2010
@hAl: Calling a test "bogus" isn't helpful. The test is written in JavaScript, and has to rely on feature detection in that language. @NielsLeenheer, @EricLaw: Let's try to determine the correct JavaScript code to reliably detect which encodings are supported by IE9 so that when someone wants to know if their video will play, they can use it.Anonymous
June 24, 2010
@hAl: let's put it this way. You ask the postman: 'can you deliver a parcel to the other side of the street?' The postman has 3 possible answers: 1 - sure, I can! I'll do it myself right away. 2 - well, maybe I could, but I'm not sure because I'm not the one who takes care of the delivery itself - I have to hand it out to someone else. 3 - I can't, sorry. Would you entrust this postman with your package if he answered 2 or 3? Here, it's the same thing - IE9 claims that it may be able, but is not sure, to decode h.264 and aac. Would you trust a browser to playback your media in that case? Well, me, I wouldn't. Considering the IE team recommends to do feature detection and to use 'every-browser' syntax, the very fact that it doesn't return proper answers on canPlayType method is akin to saying, IE9 doesn't support h.264 nor aac etc. as far as cross-browser feature detection is concerned. It's a bug that must be corrected.Anonymous
June 24, 2010
o/ Yeeeeeeeeeah! but: www.css3.info/.../multi-column-layout Please please please please please! It's in the MS Office for over a decade. Rip it off and make it work in the IE9 please! FF and Chrome already supports this and now the Opera is likely to implement it. Hundred times more important feature request than any multi-background stuff. Please please please please please! Thanks!Anonymous
June 24, 2010
@hAI: canPlayType is explicitly there to test codec support and there are very specific rules in the specification when a browser should report what. IE9 does not follow those rules at the moment and the result is that codec detection using canPlayType is impossible at the moment. IE will report the same thing for a codec that it does support and a codec that it doesn't support. IE isn't the only one with this problem, the iPhone has it too, but that doesn't make any less of a problem for users of IE. You are absolutely right that an error in canPlayType isn't the same as not supporting the actual codec itself, but then again, if the browser does not correctly tells what it supports, you can't complain about wrong test results.Anonymous
June 24, 2010
@Guarav: Agreed. I'm not here to complain about IE. I think IE9 will be a great release. Initially I just filed a bug at Microsoft Connect about this and hoped for a quick solution. But given that there were some comments here stating that the problem was with the html5test, I had to try the set the record straight.Anonymous
June 24, 2010
@Jesse Mohrland I will try to upgrade the drivers and let you know if it works. Has to be this though, since Firefox crashes as well with Direct2D enabled. ThanksAnonymous
June 24, 2010
@Niels But why do you work around the bug for the iPhone? Surely, by definition, a standards test should not include workarounds of this nature, because if a browser behaves incorrectly then it is not conforming to the standard?Anonymous
June 24, 2010
@Amtiskaw Agreed. At the time it seemed sensible to add a workaround for this specific problem on the iPhone because which codecs to support is not defined in the standard and while the iPhone is wrong, is it actually consistent in its behavior and you are able to reliably test which codec is supported. Now that IE exhibits a similar problem I am considering removing the workaround for the iPhone. Proper support of a codec not only means being able to decode video frames, but also that you properly report back if that codec is supported. If IE fails, so should the iPhone. What I can't do is simply add another workaround for IE. Because unlike the iPhone you can't reliably test which codec is supported in IE. Both codec that are supported and codecs that aren't supported give the same result.Anonymous
June 24, 2010
The comment has been removedAnonymous
June 24, 2010
hi anyone can help how can i download internet explorer 9 iam using xp i already got internet explorer 8 i tried to downl;oad the newer one but the system wont let me ?Anonymous
June 24, 2010
As far as I'm concerned, 83/100 in Acid3 test is by far the IE9 team's most important achievement. Everything else is just icing on the cake. I truly hope you'll get to 100/100 before releasing the product.Anonymous
June 24, 2010
I agree with Kroc Camen , netbook optimisations are needed.Anonymous
June 24, 2010
This is great progress! Congrats to the whole IE team! One feedback: on the video panaroma test, the animation works smoother on IE9 over Chrome. But, the movie plays smoother on Chrome. On IE9, the movie frame was skipping and slow, while on Chrome it runs smooth. You should look into that. I have a pretty good h/w and internet connection, so that's the not issue.Anonymous
June 24, 2010
Mitch, per the spec, the "postman" cannot say "sure I can." Valid responses are "probably", "maybe", and "" (no). There's a reason that there's no "yes."Anonymous
June 25, 2010
adem >> XP is not supported. Direct2D, DirectWrite, and all the cool stuff, aren't supported on XP. Besides, XP will soon be 10 years old. You should upgrade :-) Michal Tatarynowicz >> That is completely wrong. Acid3 only tests 100 features - it doesn't mean anything. The rest is much, much more important. Besides, like Mozilla, MS won't be implementing SVG Fonts, so no 100/100.Anonymous
June 25, 2010
@Michal Tatarynowicz ACID3 is fairly irrelevant as a standaard support test as quite a view of its subtests test for a lot non-relevant rare situations. Support for more dom level events, css3 element and html5 features is more usefull than spending time on some of the more less relevant acid3 subtests.Anonymous
June 25, 2010
@EricLaw Are the native codecs that IE9 has, the same as the DirectShow codecs that are standard on Windows 7 or are these different (for instance optimized for speed)Anonymous
June 25, 2010
The comment has been removedAnonymous
June 25, 2010
I just published an article about detecting HTML5 codecs and the problems you will encounter. The bug in Internet Explorer 9 I mentioned earlier is featured, but IE is hardly alone. All browsers so far - except Opera - make mistakes. Unfortunately they don't all make the same mistakes, so detecting codec support is still unreliable. While some mistakes are not that serious, some are, including the iPhone and IE9. Read the article at: rakaz.nl/.../problems-with-html5-video-codec-detection.html Cheers, Niels Leenheer html5test.comAnonymous
June 25, 2010
The comment has been removedAnonymous
June 25, 2010
The comment has been removedAnonymous
June 25, 2010
Hi am facing the same problem as Jose. But the driver seems up to date. Any other recomendation. It seems that my windows live messenger Beta also crashes due to same error. And right when I was trying give my wife a demo of how amazing it is. Any recommendations?Anonymous
June 25, 2010
btw your search is broken - it searches all across MSDN to find unrelated articles to the IE Blog that I'm trying to search! please replace it with a Google search. No one wants an MSN, MSDN, or Bing search, just use Google so that we can get the correct results - don't try to "decide" for us what we want.Anonymous
June 25, 2010
The comment has been removedAnonymous
June 25, 2010
I'm truly excited that the canvas element has been added. ...but I'm a bit surprised about the performance, it really performs wonderfully in the demos which uses a few elements but redraws the entire screen every frame, but in my own testing, rendering many simple small shapes hurts performance very badly, text and clipping hurts the performance something fierce. Is this something that will be improved upon before the final release or is the performance one can expect? Other browsers such as Chrome/Firefox can render huge amounts of elements very quickly, but performance hurts very badly from larger viewports. Which is kind of funny, because all browsers might actually implement the specification correctly, but due to the large differences in how they perform in various scenarios... might end up making it not so universal in the end, as optimizing it for one browser means that it might perform considerably worse on another. IE9 vs Chrome would seem to perform opposite of the other in most scenarios. On another note, is there any API for actually rendering in tune with the monitor or determining the HZ of the monitor? Aka, how is one supposed to render just ONCE per frame. Using a timer with 1ms delay works, but isn't really optimal and on chrome means you'll be rendering at 250 FPS which is not great. Will cleartype/subpixel rendering ever be an option when rendering using the canvas? It would seem like a natural thing, being able to render text that looks gorgeous, while also being able to render antialiased text that can be saved to an image. Instead of just being able to render blurry antialiased text.Anonymous
June 25, 2010
@search is broken You must be new to the Internet thing. If you want to use google to search this blog, simply enter site:blogs.msdn.com/b/ie after your search text in google.Anonymous
June 25, 2010
response@search is broken []just use Google so that we can get the correct results - don't try to "decide" for us what we want.[/] Nice one.... Thanks....Anonymous
June 25, 2010
anyone know why this is so much slower in IE9pp3 than in other browsers? www.scale18.com/canvas2.html compared to firefox 3.6.4 this is 20x slower on IE9pp3 (mandelbrot, 3 colors, 64 shades per color, 160x100, default zoom), and even slower on higher resolutions, while other major browsers take atmost 2x the time of firefox. otherwise, thanks a lot for increasing standards compliance of IE!Anonymous
June 25, 2010
@Jesse Mohrland [MSFT] Sure, Laptop OEM do provide drivers - but these aren't updated in a reasonable timeframe as you know. I'm pretty sure you are aware of support.amd.com/.../737-28041SupportforATIMobility.aspx for example. And yes, it's a Thinkpad R500 i've been testing on. Which one out of the following list to choose to speed up IE9 and not to loose possible "special functions" implemented by the OEM? catalog.update.microsoft.com/.../Search.aspxAnonymous
June 25, 2010
@Andy, you are the ignorant troll here, and you even posted after your "last comment on this post", you are such a loser lol. You are the one who should really stop after your declaration of "last comment on this post", but obviously you are too pathetic that you can't even keep up your own word. And your customers are all losers if they use that pathetic Chrome Frame plugin instead of Google Chrome itself, and you are the big joke if you tell your customers to use Chrome Frame instead of Google Chrome proper. If you want to use some specific APIs, go ahead, but it's just ridiculous to recommend people to use Chrome Frame instead of just switching to Firefox or Google Chrome proper. You are obviously a moron to do that, lol.Anonymous
June 25, 2010
Joed: :D You're too funny!Anonymous
June 25, 2010
Anyone else get the Error -2145124329 prevented the installation of KB2120976?Anonymous
June 26, 2010
@Doug This error means the update doesn't apply to your system. You might get this trying to install on Server 2008 which at this time is not supported for Platform Preview 3. @Ottmar Freudenberger I would recommend starting with the latest driver from the OEM if you haven't already. Anything else would just be exploratory.Anonymous
June 26, 2010
100 ms away from the market leaders! Way to go! BTW Hows about buzzing the people at HTML5Readiness.com? Last I checked it still showed IE9 as a browser that doesn't support <canvas>.Anonymous
June 26, 2010
is this a bug? www.freeuploadimages.org/.../wt6s6gh3cw049up0uce.png or this? www.freeuploadimages.org/.../deszodkfu3ptthcu0fl.png after i clicked repeatedly, the tiles have gone overlapped. is this the expected behavior? Safari does the same thing, then i think the page have a bug.Anonymous
June 26, 2010
Joed: Yes, I mainly recommend Google Chrome to my clients. Thank you for caring.Anonymous
June 27, 2010
The comment has been removedAnonymous
June 27, 2010
EricLaw, right. Indeed, one can safely replace .getElementsByName('element_name') with .querySelector('[name="element_name"]'). It's a solution that works on all main browsers. This still remains a cross-browser compatibility issue. Also (in the hope some IE developers are still reading comments from this post), is there a plan to fix the infamously broken select element's innerHTML? (support.microsoft.com/default.aspx)Anonymous
June 27, 2010
connect.microsoft.com/.../defining-options-for-the-select-element-using-the-innerhtml-method-still-fails-using-ie-9-0-third-platform-previewAnonymous
June 27, 2010
@Jesse Mohrland [MSFT] I did of course tried and did have the latest driver available from Lenovo for the Radeon HD 3400 installed. That's the one that failed with KB2028560 beeing installed and let IE9 P3 crash. The older, previous version of the driver doesn't leed to a crash in IE9 P3 with KB2028560 installed - but looking at the frame rates, it does seem to be slower in preformance than the actual one without KB2028560 beeing installed. Weired.Anonymous
June 27, 2010
echo @giuseppe: "thanks a lot for increasing standards compliance of IE!" IETEAM:
- please stop using "Same Markup" slogan, use "Standard Complaint"
- I know you have PR guys/gals at Microsoft that must do their thing, but please stop the FPS Hardware Accelerated trumpet blowing and devote more time to making IE support more standards. [see next point]
- Really good job on Canvas, i feel it brought you a lot of well deserved media attention. Now i really feel that IE team is serious and i see a change at Microsoft. Hope this new guard raise in rank. Can a leopard change its spots?
Anonymous
June 28, 2010
The comment has been removedAnonymous
June 28, 2010
@another_anon: That's simply how Windows works. In IE's case, we start the new window in the size of the last closed window. If you want to make IE start maximized every time, create a shortcut to IExplore.exe and in the shortcut properties, choose "Start Maximized".Anonymous
June 28, 2010
@another_anon: Going along with EricLaw, that is just how Windows works (and I like that). In fact, you're wrong about Chrome, Firefox, and Opera. On my Win7 machine, they all open in whatever the size of the last closed window was. And if you ask me, that's the way I want it.Anonymous
June 28, 2010
So I took the Canvas for a spin around the bock. NOT BAD! I blogged the details here: weblog.bocoup.com/canvas-in-ie9-too-good-to-be-trueAnonymous
June 28, 2010
I fail to install on Windows 7 Proffesional x64. It just stops on the first of the 3 updates required to install, and when I try to install them manually I get an error stating that the updates are not for my computer... Is there anything I can do to get ie9 platform preeview running?Anonymous
June 28, 2010
Please add to IE conditional comment <!--[if SVG]> … <[endif]-->Anonymous
June 29, 2010
I reported a crash above, it is now solved. Updated the graphic drivers as suggested. Thanks.Anonymous
June 29, 2010
Any chance you could get Google to update YouTube's HTML5 page to say that IE9PP works too?Anonymous
June 30, 2010
The comment has been removedAnonymous
July 01, 2010
@PaulT: "Well... it's great that IE9 is trying to make the web experience richer, but WHY isn't the IE9 team focussed on meeting ALL HTML5 and ALL CSS3 specifications FULLY, before adding the bells and whistles ?" x2, agree 100%Anonymous
July 01, 2010
Status of HTML5 specification: unfinished. Status of CSS3 specification: unfinished. "Bells and Whistles" being added by the IE team? None, unless you include "Using hardware so the HTML5 and css3 features actually run fast enough to be useful." Come on, do some homework before running your mouth.Anonymous
July 01, 2010
You do "some" job but this is stil low peformance graphics for today standards. On may lap (i7 with gt330) I have 25fps in 250 fishes tank, do you recommended to use desktop and gtx480 to have smooth animations or you made browser for next decades laps? It's a same you are OS producer and can't made optimized gpu powered software. Don't look in open source and other browsers just look in games powered with the same technologies and you'll see whare are you now. And other technologies like webkit and webgl sholud be implemented.Anonymous
July 01, 2010
@goxy: I'm not sure why you're not seeing higher performance. On my 18 month old Lenovo x200 laptop with a last-gen Core2 processor and the integrated Intel GMA X4500HD, I get about 41 FPS with 250 fish.Anonymous
July 01, 2010
The comment has been removedAnonymous
July 02, 2010
The "resolution" shown on the right-hand-side of the test case shows the pixel size of the browser window, not the pixel size of the screen. "Full HD" is a misleading term, since even if you mean "1920x1080", the content area of the browser window isn't that large.Anonymous
July 02, 2010
@EricLaw [MSFT] Why browser window is not in real screen resolution, probably because this "GPU acceleration" is only marketing term, this is still slow peformance browser and nothing else. When my lap powered with i7 on 2.8 ghz, DDR 3 on 1.6Ghz and gt330m with 1GB can not render simple "GPU powered" animation with more than 25fps in 1280x616 pixels I know problems are in programers that made this "animation". I had better peformances in games on intel i750 GPU 10 years ago than this engine end brand new lap.Anonymous
July 02, 2010
@Goxy: You are using the IE9 Platform Preview build, right? The browser window isn't the same size as your screen because there's a title bar, menu bar, and status bar that take away space. Thus, the content area of the browser is smaller than the content area of the screen. I'm afraid I have no idea why my much older and slower laptop runs the demos much faster than your newer laptop which has much better hardware. I can only assume that you've got something else misconfigured on your laptop (e.g. hardware acceleration disabled, etc). You'll probably want to check your task manager to see what your CPU is up to, and ensure that you're running the latest drivers for your graphics card.Anonymous
July 02, 2010
The comment has been removedAnonymous
July 02, 2010
I don't think problem is in IE9 PP,but really somewhere else,otherwise my Core i7 920 with GTS 250 would outperform i5 520M with NVS 5100M. Either drivers (apparently I never got to install beta driver... ) or Direct2D...Anonymous
July 05, 2010
@EricLaw [MSFT] and Klimax: Just look in IMDb Video Panorama example on IE9 Test page, look in picture and title text and try to move from one to another one, text is unreadable on even slow speed and pictures are with blur whatever CPU and GPU you are using, I try this on several PCs. IE9 has real limitation in object movement, probably reason is slow drawing engine.Anonymous
July 05, 2010
Goxy,I don't see problem. I can red them until it moves too fast... (all the time 60fps - used nVidia NVS)