HTML5 and Same Markup: Second IE9 Platform Preview Available for Developers

When we started planning IE9, we recognized the need for a better feedback loop with developers. The developer community was clear that they wanted pre-release builds of the browser platform in a consistent rhythm, with a good feedback mechanism.

Seven weeks ago at the MIX Conference, we released the first IE9 Platform Preview. We committed to updating the Preview approximately every eight weeks. Today, we’re releasing the second Platform Preview of Internet Explorer 9, available now at

Today’s release builds on the first Platform Preview, delivering improvements to IE9’s performance, support for standards, and hardware acceleration of HTML5.  We’ve also updated the test drive site with a new set of developer samples to show what developers can do with GPU-powered HTML5. As part of our commitment to enabling developers to use the Same Markup – the same HTML, CSS, and script – on the web, we have contributed many new tests to the W3C for HTML5, as well as CSS3 Media Queries and DOM. The Developer Tools in this preview include some new features to make finding and fixing markup issues easier.

Developers should expect much more from browsers in order to deliver the graphically rich, interactive applications that HTML5 will enable. In IE9, our goal is to provide professional-grade, modern HTML5 support on top of modern hardware through Windows. The IE9 Platform Preview and the samples at the test drive site show the significant performance gains that web pages enjoy when a browser takes full advantage of the PC’s hardware capabilities through the operating system.

Performance and GPU-Powered HTML5

With the second Platform Preview we continue to improve IE9’s performance and maintain our focus on real-world sites and hardware acceleration. We examined the patterns in use across many websites and frameworks to identify which changes in the browser make actual sites faster. The first chart here, for example, shows how different subsystems contribute to the performance of different real-world sites.

The JavaScript engine is one of several important subsystems. Here’s a chart of IE9 performance on one particular industry benchmark for JavaScript performance, the Webkit Sunspider test:

Webkit Sunspider Javascript Benchmark Results from 5/5/10. IE9 Platform Preview #2 is competive with other browsers.

You’ll notice that the performance difference between IE9 and other browsers on this benchmark is in the range of an eye-blink. As we continue to make IE9’s script engine faster for real world sites, IE will continue to become faster at this particular benchmark as well. To date we’ve done very little specific tuning for Webkit Sunspider. As with most benchmarks, depending on your machine, the differences may vary.

The performance you experience browsing actual websites often has less to do with JavaScript than with other subsystems in the browser. You can see this for yourself by trying the Speed Demos at the test drive site in IE9 and in other browsers. This video shows them in action:


link to video (.wmv) , (.mp4)

These samples show just some of the advantages of GPU-Powered HTML5, and applying the power of your PC’s hardware to making web browsing faster. What’s particularly exciting is that developers don’t have to rewrite their sites – their current markup (HTML, CSS, and script) just runs faster.

Same Markup

Web browsers should render the same markup – the same HTML, same CSS, and same script –the same way. That’s simply not the case today. Enabling the same markup to work the same across different browsers is as crucial for HTML5’s success as performance.

While some people associate this same markup situation with IE6, it actually applies across browsers in general. Developers typically have to write different markup to get the same desired outcome, even across the latest versions of Firefox, Chrome, and Safari browsers.  You can see some examples in these videos with people on the IE team about DOM, SVG, and CSS3. On the test drive site, you can try the “Border Radius” example in different browsers to see another. There are many examples today of developers trying to use HTML5 features, only to find it works in Safari only, Firefox only, or Chrome but not Safari or Firefox (find ‘doesn't work well on Firefox’ on the page).

We’re engaged with the standards working groups and other browser vendors as part of the web community with Same Markup as a key goal. Same markup is the real-world benefit of standards for developers (and through them, the rest of the web).

True standards bodies are important so that different parties and communities can come together to consensus. Specifications are a good start. Other parts of the technology industry have shown that comprehensive and accurate test suites are essential to provide interoperability for developers and products that work for consumers. They’re how an industry converges on a common understanding of the specifications and can assess its progress.

Enabling an interoperable web so developers can create amazing HTML5 applications is at the core of what “same markup” means. 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.  Today, we’re submitting 88 new tests to the W3C bringing our total to 192 tests submitted during IE9.  Our focus leading the HTML5 Working Group’s testing task force is on thorough, professional-grade tests that enable developers.

On the test drive site, you can try out several examples of IE9’s improved standards support and interoperability with other browsers. CSS3 Media Query support enables a site to adapt to different display attributes (like width, height, orientation) on the fly.

Optimized for Large Display Optimized for Netbook Optimized for Mobile
CSS3 Media queries demo displaying in large screen format CSS3 Media Queries Demo displaying in Netbook format CSS3 Media queries demo displaying in mobile format  

Another sample on the test drive site shows how developers can use DOMContentLoaded to deliver better performance with web pages that respond to the user as soon as the page is done parsing, instead of waiting for when everything on the page has loaded.

CSS3 Media Queries and DOMContentLoaded are examples of requests from the community that are now in the product. Many other features you’ve requested  – DOM Traversal and Range, getElementsByClassName, createDocument, and more – are available today in the IE9 platform.  The Preview also fixes many issues that users have reported with our SVG, JavaScript, and CSS3 implementations.

Some people use Acid3 as a shorthand for standards. Acid3 tests about 100 details of a dozen different technologies. Some are still in “under construction.” Some of the patterns, like SMIL animations, are inconsistent with other parts of HTML5, like CSS3 animations, and need to be reconciled. Here’s a screenshot of how today’s IE9 Platform Preview runs today’s Acid3 test:

Acid3 test - 68/100

As IE runs more of the “same markup” that developers actually use on the web, our Acid3 score will continue to go up. This approach is a partial test of some capabilities, and is quite different from a test suite. For contrast, you can run some of the tests we’ve submitted to the W3C here in any browser.

The Developer Tools in IE9 Platform Preview 2 include new features. The Console window is now a full tab that includes diagnostic information from IE.  Developers can use the “Change User Agent String” menu item to experiment with sending different that UA strings to sites with every request, selecting from preset strings or creating their own custom string.  This complements another feature we’ve included – the new IE9 UA string.


IE9 Dev tools new User Agen String switching tool

Since releasing the first Platform Preview, we’ve answered a lot of developer questions. Here’s a recent one from a GMail engineer:

Sometimes we have to do browser detection, and we tend to group all IEs together, with the occasional branch based on version.  Should we treat IE9 entirely differently?  Is it intended to behave closer to webkit/gecko than IE7/8?

Our answer is an emphatic YES. IE9’s standards support makes it much closer, for developers, to Gecko, Presto, and the different versions of Webkit than it is to IE7. We want to make the same markup work across browsers, and want feedback from developers about the issues they find offering the same markup to IE9. To this end, we’ve shared some suggestions in this blog post on Same Markup: Writing Cross-Browser Code and will blog about additional examples and patterns.

We recently wrote about HTML5 video and codecs. HTML5 Video tag support will come in Platform Preview 3. We want to focus with the community on the underlying and enabling technologies, like CSS and DOM, first, so that the same markup really does work across sites.

Test Drive It Today

We’re eager for developer feedback, and hope you have as much fun test driving IE9 as we continue to have in building it. We appreciate your support and interest in IE and will continue to pay close attention to all the feedback on Connect.  We’ll continue the rhythm of IE9 Platform Preview updates approximately every eight weeks, as well as our work at the standards bodies with other browser vendors and the community. Your feedback about how IE9 handles the same markup your sites give other browsers, and the standards tests we’ve submitted to the W3C, makes a huge difference.

Many people have asked about the beta and the full browsing experience. We want the beta to build on a solid platform of performance, same markup, and GPU-Powered HTML5. Our focus now is building that solid platform.

A few last details: the Platform Preview continues to be the thinnest possible wrapper around the web platform, and as such is not intended for general purpose browsing. It does not have IE8’s security protections (like Protected Mode, SmartScreen filter, and the Cross-site scripting filter) which we strongly recommend for anyone browsing the web today.

The platform preview installs side by side with Internet Explorer 8 so that you can try it without replacing the full IE that comes with Windows. This second release of the Internet Explorer 9 Platform Preview will install over the first version. There is no need to uninstall the first Preview before installing second.  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

Edit 2:20pm – updating links to the standalone versions of the embedded video.

Comments (156)
  1. Anonymous says:

    mind blocking affects,usage and many more with HTML5 thanQ u for the fast release…

  2. Anonymous says:

    Got 317.4ms in Opera 10.53 on Windows 7. IE9 sux.

  3. Anonymous says:

    I would test this however…

    1.) It’s not available on Windows XP. Anyone with half a functioning brain knows you guys can get this to work without the GPU functionality…heck we know you guys could get the GPU functionality to work in XP.

    2.) My upgrade edition of Vista (which hogs up a full GB in VirtualBox) expired and won’t ask me for a previous Windows CD and/or key for a previous version of Windows. I refuse to use it or buy Windows 7 as they SEVERELY impair my ability to work efficiently.

    Guess I’ll test this out when I have extra time to uh, reinstall Vista, reinstall Vista SP1, reinstall Vista SP2, reinstall that Direct2D patch, etc etc. Let’s see, when I did this before the first developer preview it took me about ten hours, great!

    All this added to the fact that you guys are only supporting a proprietary video codec for HTML5 while you’ve made it clear that you’re covering the widest range of actual standards?

    Yeah, my enthusiasm has really been shot. Don’t get me wrong, keep up the great work adding actual standards! Though as excited as I am to see IE become standards compliant I am so beyond having my time wasted especially after having yesterday entirely blown by an incompetent web host who can’t even have PHP sessions saved to a file among other idiocies. Time is worth money and I’m not sitting on an empire of money where the sun never sets.

    Again congratulations (and I mean that) though you guys keep boxing yourselves in so much that you’re negating all the positive vibes you should be getting.

  4. Anonymous says:

    toi thay ie9 nay rat co chien vong co the co rat nhieu nguoi ua chuong

  5. Anonymous says:

    Holy crap!

    This is the first time I haven’t wanted to leave a negative comment here.

    Fantastic work guys, please please keep it up.

  6. Anonymous says:

    You’re losing more and more credibility every time you try to pull crap like this. Everyone reading this blog knows how dishonest you are by comparing a hardware accelerated version of IE to the non hardware accelerated version of Firefox and Google Chrome. Please, give us some credit! Redeem yourself by redoing the video with hardware accelerated Firefox!

  7. anonymous says:

    XP support? Chrome 5.0 Beta runs on XP. Chrome 6 is even planning to add hardware acceleration. If they use a cross platform technology like OpenGL then IE9 will once again end up not supporting XP for no reason.

  8. @Dean Hachamovitch

    > our commitment to enabling developers to use the Same Markup – the same HTML, CSS, and script – on the web

    > Web browsers should render the same markup – the same HTML, same CSS, and same script –the same way.

    You definitely misuse markup in those sentences.

    And again, you assume that the same markup, same HTML code – no matter how broken, invalid and malformed it can be – should be rendered the same by web browsers. It is not the case! In quirks mode, it’s anyone’s guess and there are several incompatibilities. While in standards compliant rendering mode with a valid doctype declaration referencing a strict DTD, invalid markup code can be rendered in various ways, in an unpredictable way. Whenever there are validation markup errors in a webpage, then there is no guarantee whatsoever in any web browser.

    Gérard Talbot

  9. greg says:

    Please allow us to crop out the left and right edges from web pages when viewing or printing.

    This is a problem when you view on a small device or print to a PDF or paper copy.  Many web sites do not render well when output that way with the left hand navigation table or right hand advertisements panel being printed before or after the main page content.

  10. TheMan says:

    Love the H/A demo.  Obviously, it blows all other browsers out of the water except for Opera; which achieves similar results w/o H/A.

  11. Sylvain Galineau [MSFT] says:

    Gerard, having spent a day at a web designer conference where everyone used ‘markup’ to describe HTML, CSS and Javascript, I can sympathize with your semantic concern 🙂 But it’s now in the culture.

    Also, Dean is not the only one to assume that invalid markup should render the  same across browsers. An important part of HTML5 is to define common parsing and error handling rules so that browsers produce the same DOM tree given the same input, whether that input is malformed or not. Given the same DOM tree, the odds of achieving the same rendering are certainly a lot higher than they are now, with each browser implementing its own HTML fixup logic by reverse-engineering the others.

    This doesn’t mean everything is going to be valid or that malformed is OK. Valid markup is still best and remains the ideal way to maximize cross-browser interop today.

    But it does mean that malformed markup is going to render more an more alike in the future. The increasing compatibility across all markup – old and new, valid or not – is definitely a good thing for users.

  12. Robin Berjon says:


    any chance of an OS X version? It’s a pain to have to boot up Windows just to test with IE!

    Keep up the good work!

  13. Rob says:

    "Developers typically have to write different markup to get the same desired outcome, even across the latest versions of Firefox, Chrome, and Safari browsers."

    Typically, no. For IE, yes!

    What does hardware acceleration have to do with web standards? Once again you’re promoting something that only works on Windows and only works in IE and only those computers that have such hardware.

  14. @Robin: Sorry, no.

  15. Damian says:


    Firefox 4 also plans to have hardware acceleration, both in Direct2D and OpenGL flavors. But they will be able to provide a lot more with Direct2D has more complete APIs for this sort of work.

  16. someone says:

    I have a C2D laptop @ 1.86 GHz, 3 GHz RAM, Intel 965 Express Graphics with WDDM 1.1 drivers, Windows 7 32-bit yet I get 938.1 ms latency in Sun Spider test. What’s going on? With Platform Preview 1 also, I got above 1000 ms. Are you posting fake results?

  17. Templarian says:

    Getting better. I really hope you guys update this JS engine though. Not even close to running the same JS version as others, which makes me question the fairness of that graph.

    Hopefully this will be WP7’s browser later on, because as it stands the IE in the emulator appears to be IE7.5 (based on testing). This is really bad when you figure Microsoft is starting over and could have just used an open source alternative like WebKit.

  18. James Tiedt says:

    The #2 IE9 Preview screams on my HP desktop. It loads sites like imediately and OMG the browser tile game is fast compared to Chrome 5 beta. getElementsByClassName is long overdue but still missing from 8. You guys have turned around a 10 year sinking torpedo. Cheers to the team. Please stay focused on IE9 this time.

  19. Blah says:

    someone: Don’t be silly. What are your numbers in the other browsers.

    templrin: <<Not even close to running the same JS version as others>> suggests you don’t understand much about JavaScript versioning.

  20. Sylvain Galineau [MSFT] says:

    @Robin, good to see you here.

    Well, if you ran Windows all the time it wouldn’t be a pain to boot it up…

  21. Glen says:

    Are there plans to add CSS3 box-shadow and text-shadow support later in the development cycle?

    These along with border-radius are commonly supported by other modern browsers, even if they require prefixes.  Adding support for these would go a long way toward universal code in my style sheets.

    BTW – Good work on the speed and compliance advancements!

  22. William J. Edney says:

    Imagine my surprise when clicking on the "JavaScript" link in this sentence:

    "The Preview also fixes many issues that users have reported with our SVG, JavaScript, and CSS3 implementations"

    sent me off to my own bug in the Connect database… 🙂

    Now off to grab a copy of the second preview… keep fixing bugs guys… looking good!


    – Bill

  23. Templarian says:

    @Blah, I’m thinking you know very little about JavaScript. They don’t even have get and set, lets not even get started listing all the other objects missing. Yes, we all know versions mean nothing when IE doesn’t add the features required to meet said version, yet have others.

  24. @Sylvain Galineau [MSFT]

    > An important part of HTML5 is to define common parsing and error handling rules so that browsers produce the same DOM tree given the same input, whether that input is malformed or not.

    That’s why it is going to take until 2022 before HTML5 becomes official, stable, reliable. In the meantime, why Microsoft and IE Team and MSDN, etc, do not emphasize, underline that valid markup code is not only necessary but a minimum? and encourage web authors to always write valid markup code? HTML4 is still going to be widely in use until 2022..

    > Valid markup is still best and remains the ideal way to maximize cross-browser interop today.

    Sylvain, you say so and you believe so with and/or without your [MSFT] decoration but it’s definitely not the case for the whole Microsoft people. Have you been visiting MSDN (category web development) recently?

    Malformed markup code, invalid markup code, validation markup errors should be discouraged by any developers+managers on the IE team… but so far, I think you’re the only one who spoke about this.

    The very first webpage ever created by Microsoft some 15 years ago failed validation and the very last webpage published by any Microsoft-controlled website still fails to pass markup validation and CSS validation. It’s even true for a lot of IE Team members.

    962 validation errors today:

    regards, Gérard

  25. I think it’s quite unfair to compare a developer preview of IE with hardware acceleration to the lastest stable release of Firefox, without hardware acceleration. The Firefox 3.7 Alpha has had hardware acceleration for a while, just flip two switches in about:config. The plan is to have it activated for the final release.

  26. Will Peavy says:

    Is there any chance a VPC image, with Vista SP2 (or newer) and IE9 Preview, can be made available?

  27. Mehmet says:

    Thanks for the update, we really appreciate these blog posts about the progress you have been making.

    I would really like to know your stance with the canvas tag.

    So far I have not seen *anything* about the canvas tag except the comments section, it’s as if you are avoiding the subject all together.

    Is it planned, if it is, which platform preview version?

    While it is possible to do svg animation, it is by nature not as suitable as a drawing api, which can be provided by canvas.

    Ithink it would be a big mistake not to support it, especially after all the good work done by the IE team on hardware acceleration.

  28. AlfonsoML says:

    What good is a browser if most of the people keeps using XP and won’t be able to use it?

    Maybe you should make some arrangement with the company that developed that OS and make it work there.

  29. > Developers typically have to write different markup to get the same desired outcome, even across the latest versions of Firefox, Chrome, and Safari browsers.

    Generally speaking, that is not my experience and it is not what I hear in web authors’ blogs and in web authoring (HTML, CSS, DOM, DHTML, javascript) newsgroups discussions. If we’re talking about some specific CSS3 properties (eg gradient) still in a module in WD or some new HTML5 elements or attributes, then maybe or quite possible.

    If such statement was generally true, then it would prove the utmost irrelevance of test suites, acid tests and many other browser conformance testings.

    Gérard Talbot

  30. Tom says:


    You can not compare IE (hardware acceleration) to Firefox/Chrome without hardware acceleration.


  31. Brian LePore says:

    Okay, maybe my coffee hasn’t kicked in, but what exactly is the difference between Browser Mode and Document Mode again?

    Looks like there needs to be some work on TinyMCE to work with IE’s new engine. Great. =/

    I don’t know if you’ll get to things like box-shadow or text-shadow, but what will you do when used along side the filter property?

  32. John says:

    Why is it that first class browsers like Chrome and Firefox are available on XP in their latest iterations but not this?

  33. Gyrobo says:

    Your scrolling text demo [1] seems like something that would be better implemented with CSS Animation and 3D transforms [2].



  34. Brian LePore says:

    Okay, maybe my coffee hasn’t kicked in, but what exactly is the difference between Browser Mode and Document Mode again?

    Looks like there needs to be some work on TinyMCE to work with IE’s new engine. Great. =/

    I don’t know if you’ll get to things like box-shadow or text-shadow, but what will you do when used along side the filter property?

  35. raffi says:


    Because IE9 is going to be released at a future date when even less people will be using XP.

  36. Gyrobo says:

    @Raffi and John

    Actually, it’s because IE9 relies on APIs for hardware acceleration that are available only on Vista and newer. Supporting Windows XP would require coding a separate graphical backend.

  37. Radek says:

    I’m not Developer, but I love this build, it’s very stable, good acid 3 score, flash integrated, only wish it had UI. We have to wait  7 – 8 weeks for a version with UI? Keep up good work.

  38. infinte says:





  39. Chess says:

    I can’t, nor would I want to, install Silverlight on my Ipad, which is what I’m using to read your site right now. Embedded video fail.

  40. raffi says:


    Yeah ipad really fails at embedded video.

  41. John says:

    I understand the desire to impress with H/A, but not supporting XP just irritates all the web developers that you are trying so hard to please. Every web developer is looking at what you are doing and immediately saying, "Looks great, but since XP can’t upgrade I’ll still be writing broken IE 7/8 code for the next ???". I don’t know how long XP will still be around, but Microsoft hasn’t end of lifed it yet, which means long enough.

  42. Jesse Mohrland [MSFT] says:

    @Brian LePore

    Browser mode dictates how conditional comments are evaluated and what UA string is sent to the server.

    Document mode dictates how markup is parsed, rendered and what OM functionality is available.

    Only document mode can be controlled by the content author while browser mode is toggled by the user and the Microsoft compatability list.

    Toggling the browser mode can affect the default document mode for a given page.

    For more info, look for [How IE8 Determines Document Mode] posted at the beginning of March.

  43. David Bruant says:

    I’m sorry to probably do again a comment that I may have already done, but in your testing center (, the "Cross-browser Test Results Summary" leads to think that IE is more standard-compliant (full dark green 100% column) than other web browsers, what most people know is currently false even for IE9.

    Where are your tests for the <canvas> element ?

    Where are your tests for the new HTML5 tags (section, article…) ?

    Where are your tests for .textContent (instead of the current IE-specific .innerText) ?

    I think that you should add a mention to say that it’s these tests are developed for IE9 development phase.

    Moreover, you only give results for the tests you submit to the W3C. Aren’t there other already tests ?

    I would like to point out also that Microsoft has developed an ECMAScript 5 test suite ( Why don’t you add it to your testing center ? I must admit that so far, you seem to only deal with DOM issues, but the testing center seems to be generic.

  44. jOHN says:

    Thanks Gyrobo for the answer.

    I hope that development at least continues for XP so that the majority of users can have the best experience possible.

  45. heavencanwait says:


    Hi.. will we see something like XNA for HTML5 some time in the future?

    And about all discussion on video… Is there any chance of that this fight turns into charges for video for the consumers? Actually, I think that’s the question. You tell us about a "standard" that companies created and can, any time in the future, start to charge us for… or not?

  46. heavencanwait says:


    Hi.. will we see something like XNA for HTML5 some time in the future?

    And about all discussion on video… Is there any chance of that this fight turns into charges for video for the consumers? Actually, I think that’s the question. You tell us about a "standard" that companies created and can, any time in the future, start to charge us for… or not?

  47. Mark Grgurev says:

    I found a standards related problem with XHTML in IE. Actually, I found it in every browser but Opera, and that’s that the XHTML specification says you can self close any element if it’s empty. Thus, I set up one of the test pages on my server to check the version number of the browser and if the version number is equal to or less than 8 it should show this:

    <script type="text/javascript" src="js.php"></script>

    Thus it will show correctly in Chrome, Firefox, Safari, and IE.

    and if the browser version is above 8, it shows this:

    <script type="text/javascript" src="js.php"/>

    The result is that Opera, which is well above it’s 8th version will display the page correctly while IE9 can’t render anything below this line thus leaving the whole page blank.

    While I could very easily tell it check for version numbers below 9 instead. I would rather this be fixed in IE9 since it’s something that is part of a standard that is now over 10 years old.

  48. Joe says:

    Dean – with these improvements it’s going to make it harder to justify the investment in Silverlight. This is sad given that the developer experience is so much better than HTML5/JavaScript.  

    Any comment on the discussions earlier this week of integrating EMCA CLI and ECMAScript from Joe Hewitt and Miguel:

  49. @David Bruant

    > Where are your tests for .textContent (instead of the current IE-specific .innerText) ?

    Please visit

    DOM 3 Core textContent is now over 6 years old. Firefox 1.5 supported textContent.

    regards, Gérard

  50. Aryeh Gregor says:

    I’m going to have to pitch in and ask about <canvas>.  Discussion by some browser implementers in #whatwg on freenode concluded that it probably had more author demand than SVG, and would be very easy to implement on top of Direct2D.  Could we please have a statement on this?  Is there a problem with Apple’s patents on it, which won’t be royalty-free until the canvas spec reaches REC?  Is there some other problem?  I’m guessing IE9 support for <canvas> isn’t totally off the table yet, because otherwise you’d probably say so, but clarification on that would be awesome.

    I personally only use open-source software (on Linux) if I can help it, but as a web developer, let me say: *thank you* for making IE work properly!  I’m really waiting to be able to use the same code paths for IE9 as non-IE browsers.

    @raffi: The iPad displays H.264 video fine using HTML5 <video>.  For instance, it should be able to handle most YouTube or Vimeo videos with no problem, if you use the HTML5 beta.  It’s only proprietary technologies like Flash and Silverlight that Apple refuses to support.  (Of course, in other ways they’re quite evil, and I stay even further away from them than from Microsoft, but I give them credit for their pro-web standards stance.)

    @Mark Grgurev: Like all other browsers, IE9 will not treat content served with a text/html MIME type as XML.  Make sure you’re serving your content with an XML MIME type, and then it should work correctly.  If Opera tolerates self-closing <script> in text/html, that’s a bug per the HTML5 spec, and I expect it will get fixed within the next couple of years as browsers switch to an HTML5 parser.

  51. Kroc Camen says:

    Excellent. Thank you for keeping everybody in the loop with the updates; there is much to digest here. CSS3 media queries is massively important addition.

    I was hoping my website would be working with this build, but the conformance issue with html/head/body elements being optional is not yet fixed. I will wait for the next release.

  52. Stu says:

    Good work on Acid 3, your way past the psycological 50% mark, and heading towards the home stretch.

    Try and hit the top 1/4 now, and get > 75/100 🙂

  53. Paul says:

    Ehhh… Install Silverlight? HTML5?

  54. Acid3 says:

    At this rate IE9 has a chance of getting 100% on Acid3 before firefox !

  55. Xael says:

    I absolutly love this build. Keep on going with this great work guys.

    For all those XP-users, wake up! We all live in 2010. What exactly do you expect from Microsoft? Supporting XP until you die? MS is first and foremost an economically company. They have to sell Windows licences to invest in new products like IE9. If you all just sit forever on XP, MS will not make any money and so no further development can be ensured.

  56. 6205 says:

    This will be frakin’ best web browser on the planet 🙂

  57. Neuro says:

    @Xael: I don’t care about XP support as an user, but I do care a lot as a developer. If 55% of our user base will be stuck with IE8 on XP then thinking "yada, you losers, wake up" is not gonna help me.

    Many people never update OS, they just buy new computers. And pretty powerful machines have been sold with XP in the last years – for regular user (mailing, some surfing, a bit of youtube, sorting photos) those will be enough for the years to come.

  58. Please MS implement the canvas tag. Web based Image/Photo editting programs rely on it. Also a program I’m working on relies on it because I need direct pixel manipulation (I’m building an HTML5 Photo Mosaic Generator).

  59. Sylvain Galineau [MSFT] says:

    @Gerard, yes, there is a lot of invalid markup out there. Here is one random example:

    I am getting 37 errors and 29 warnings using the W3C’s HTML Validation service at If I force a check as HTML 4.01 Transitional, I still get 30 errors and 38 warnings . OK, maybe it was not that random a choice 🙂 But it’s one interesting data point among, probably, billions that would indicate that standardizing parsing so that all markup renders the same across browsers is not just a very reasonable goal, it is the only sane long-term option. It just is in no one’s interest, least of all browser vendors, to compete on proprietary error handling.

    Now, don’t  get me wrong: I’d love to fix all these errors and do it yesterday. But allow me to be candid: if I have to rank all the things the IE team ought to do by priority, editing and re-testing hundreds of web pages – which will look exactly the same to end users once we’re done – does come after ensuring their content is useful and accurate, and in my own job that even comes after making sure we fix the CSS bugs you report and pass the great testcases you submit, which is part of improving interop across the standards we support, which includes tasks such as representing Microsoft on W3C Working Groups, and all those things are ultimately about shipping IE.

    And I suspect that’s pretty much the same everywhere – maybe even at Mozilla… – hence the prevalence of invalid markup on the web. I just don’t believe the latter proves us all to be bad, careless, incompetent or lazy people. Now, I know that from your vantage point Microsoft has infinite resources. Oddly enough, *we* feel like there is never enough time.

    But even if we did have infinite resources, I still find that letting software deserialize and normalize document trees according to standard rules would be a more effective solution than fixing billions of past, present and future pages which in turn implies fixing the billions of lines of code and scripts that generate them in part or in whole. If Microsoft, Mozilla, Apple, Google and Opera agreed on one browser slogan, I’d actually be willing to bet on "Same Markup, Same DOM".

    Malformed markup will simply happen. Content is aggregated, mashed-up, user-generated and generally manipulated in too many ways to guarantee validity for any length of time beyond the most static of pages. Giving end-users the web equivalent of the blue screen of death every time the tiniest bit of invalidity occurs – as XML and thus XHTML require – is just not a viable option. We *must* handle malformed markup. The same way. Across browsers.

    Maybe you think allowing browsers to handle this massive soup of broken HTML is a huge bug. Some of us believe this is one of the central features of the web’s success. Standardizing this feature and implementing it is something nearly every single browser vendor I know of is working on at one level or another today.

    This being said: if you do run into invalid markup on our sites that does make a browser choke – as in broken content or functionality – you’re welcome to let me know and I’ll take care of it. Thanks !

  60. Claus says:

    There were few things in the world I hated more than the MSIE team for what they’ve done to my poor little soul for years and years and years…, well. However, that seems to be a story of the past.

    Big congratulation for what seems to become the best product ever shipped by the IE team. Awesome benchmarks, promising HTML5 support. Wow! Keep it up!

  61. Paul says:

    Awesome. By the time IE9 ships you should only be 50% slower than the then current version of Chrome or Opera.

  62. Jon says:


    I can’t find the source, but I remember reading that someone from Intel mentioned that they knew Microsoft was working on a Canvas implementation on top of Direct2D. And there have been remarks from other MS employees on this blog and elsewhere that have hinted that it will be supported. However, I wouldn’t expect to hear an official announcement until it’s actually available in a future developer preview. And as you said, there are potentially some legal difficulties with Apple that might delay it.

    Personally, I’d say don’t get too hung on Canvas. It’s popular because it offers an easy, immediate graphics model that many developers are familiar with and can easily work with. However, on the web, where sceen size and resolution varies massively, a retained, vector based graphics technology like SVG is far more appropriate. It’s not as easy to work with of course, but in the long run you’d be far better getting your head into SVG rather than waiting around for Canvas IMO.

  63. Where is <canvas>?

    If you don’t implement it, MS is not real about HTML5!

  64. dlh2009 says:

    Keep up the good work guys. I am extremely impressed with the huge jump on the Acid3 test score. 68/100 is very impressive. Just think you are only 32 points away from a 100.

    If you guys can live up to the hype that IE9 is made out to be, you guys will be guaranteed that some Firefox and Chrome users will return to IE.

    Now, if you guys can increase the security in IE9 without slowing it down, I will be extremely impressed.

    Keep it up!

  65. Steve says:

    Hi Are you going to have a Mac version? And sorry if I missed this… When is the target date of this being done?

  66. Love the IE Previews. Is there any way to make IE8 use the rendering system from IE9 (Trident 5, the new JS system…)? Maybe something like ChromeFrame does?

  67. Steve says:

    I think I heard IE 9 will only work with windows 7…

    So how in the heck do you test for this if it only works for one specific operating system…  Are you forcing all freelance web designers to go out and buy windows 7 even if they are using another platform just to test their websites?   I don’t want to buy a whole OS just for browser software.


  68. Vlad says:

    Small fonts still look noticeably worse than cleartype.

    Try reading email or using Google Reader with it. The text looks edgy and angular and spidery.

    I don’t know whos idea it was to focus on having fonts scale really huge with nice edges when the vast majority of browser use is at small font sizes.

  69. Ray Stantz says:

    Can you provide an Ogg Theora version of the video? It will make playback in Firefox very convenient. Thanks.

  70. EricLaw [MSFT] says:

    @Steve: IE9 will run on Windows Vista and later versions. To run it on a Mac, you’ll need to use a Virtual PC / Parallels / etc.

  71. Matt says:

    Paul: Sunspider is one benchmark, and one that measures only one thing. The real web is about much more than what Sunspider tests. Try running the IETestDrive site’s demos in Chrome and I think you’ll see that the Google guys have a lot of catching up to do.

  72. Badger says:


    No, that’s not cheating. H/A is a feature which IE9 supports that others don’t. If you don’t let that count, then you might as well say that saying "IE8’s JS engine is slow compared to others" is CHEATING because the other JS engines are JIT-ted. IE8 doesn’t support JITting.

    @ everyone who wants canvas

    Well, I do too actually. But, at times it does seem like there are two standards (SVG & Canvas) competing for similar functionality. I realize their differences in graphics methodologies but the end results are very similar. Yet if MSFT only has the time to commit to one of them, then I’d much rather them pick SVG.

  73. GmailNeedsClippy says:

    Since Dean mentioned they received a question from a Gmail engineer about differentiating content, does that mean Google is working on an IE9 Standards Mode version?  Hopefully.

    Because right now it loads in IE9 SM but is not functional. Switching to IE8 SM, it does load correctly.

  74. infinte says:

    You have made EVERYTHING in IE9 DirectX faces and textures, Why not a direct manipulation on it? Why no more direct manipuoation on these faces and textures?

  75. Wurst says:

    Great job!

    But the Flickr demo needs some mip mapping.

  76. raffi says:

    Is there anyway to force software rendering?  I ran IE9 on a slow machine, Win7 and ATI Radeon x600 and hardware acceleration seems to make performance worse.  I think it’s because there aren’t Win7 video drivers provided for old DX9 cards by ATI.  So it’s using some slow Windows OOB driver.  Or maybe the card is just terrible.

  77. Hamranhansenhansen says:

    Sounds like Microsoft is finally making a Web browser! Looking forward to it.

  78. EricLaw [MSFT] says:

    @Raffi, yes, you can force D2D mode; search the release notes for "D2D Force" at

    Having said that, an x600 should be plenty fast– it’s definitely faster than the card I have in this machine, which runs the PPB demos pretty quickly.

  79. Frank Olivier [MS] says:

    @raffi We’re checking IE performance on a wide variety of hardware; where we find issues, we will fix them. 🙂

  80. Wurst says:

    @Damian: Actually, more D3D than D2D according to their goals.

    @infinte: Managed code is not suitable for such an application.

    @Aryeh Gregor: AFAIK, only H.264 Baseline videos, that’s why Google had to recode some videos.

    @Xael: Well, one might come up with the fact that Visual Studio 2010 and Office 2010 still support XP…

    @Dominic Pettifer: Subpixel positioning might pose a problem (like it did in WPF and they made it toggleable in WPF4)

    @Jon: I aggree. SVG > Canvas.

    @Steve: Vista too.

    @Ray Stantz: Please read the previous 2 blog posts, they explain that MS – unlike Google – is afraid of the legal situation surrounding Theora/Vorbis. The don’t believe the devs from ON2 and Xiph where good enough at avoiding patents they haven’t licensed.

    BTW: The "DOM Range & Selection" demo is broken?

  81. Wurst says:

    @Frank Olivier[MS]: Including on PC with a with a VIA processor and a Matrox GPU? 😉

  82. Matt says:

    As many other web developers, I’m glad Microsoft has given a green light for the IE team to embrace web standards and improve inter-browser compatibility. That certainly makes our jobs much easier.

    I’ve been developing for the last two years a web-based rights violation database. When the project started, I made a deliberate choice not to spend time hacking code to make it work on IE6/7/8.

    The direction IE9 is taking made me reevaluate the decision. After downloading the platform preview, I’m happy to report it _would_ work flawlessly if one more IE bug is fixed: an improved getElementsByName() function that captures all elements (a behavior found in firefox, opera and webkit-based browsers).

    Can the developers confirm this will be fixed in final IE9?


  83. Mark Grgurev says:

    …I had already declared an xml version thus I assumed IE would default to excepting it as application/xhtml+xml. Just to make sure though, I added the appropriate meta tag as well and IE8 and 9 will still not render the page.

    This is the test page I keep talking about.

    If you look at the source code, you’ll see that nothing passed the self-closed script tag will work, not even the title tag.

  84. OK, it’s better than the last preview (for example 68 instead of 55points at the acid-3 test), but there is still a lot of things to do, to call it a modern webbrowser!

  85. Markus says:


    You should join Microsoft Connect and file an incident for the desired behaviour if it dosn’t exist yet. Requests in comments are more easily missed than requests in bug tracking systems.

  86. To those saying we don’t need Canvas because we have SVG. The two aren’t competing technologies, they are meant for different purposes.

    SVG allows scalable resolution independent images with animation, and it gives you a DOM to play with in JavaScript, good for UIs.

    Canvas provides direct pixel manipulation and is suitable for drawing/paint applications and direct photo editing and photo manipulation (think correcting red-eye, airbrushing, that sort of thing).

    Really we need both technologies, otherwise MS will just show that they’re behind the curve yet again (all the other browsers support Canvas).

    @Wurst: I wasn’t referring to subpixel manipulation, don’t know where you got that from. But would SVG allow me to load up an image then look directly at it’s individual pixels and modify them? I could be wrong but I don;t think it does.

  87. oliver says:

    For getting more test audience, maybe you could add the IE9 preview to

  88. izdelava spletnih strani says:

    I’m already a fan! The video is great, but i imagine that those demos were ‘written’ for IE9. Anyway if you sum up + and – i’m convicned that IE is still the best browser out there.

  89. Jonas says:

    The improvements are impressive. However, is implementing a basic UI with an address bar and back button so difficult?

  90. kabkab says:


    re Gerards excellent comment at the top.

    I guess you can believe it’s in the culture if it helps you sleep at night.

    I’ve never heard of CSS being referred to as markup until I read this blog post.

    It seemed strange to me because I have never had to write different markup to get the same result cross browser.

    The only thing I’ve had to do different is the CSS for IE.

    CSS3 spec isn’t final, CSS2.1 has been around for ages and you guys have still not got it right, so pointing out problems between Firefox Safari and Chrome makes MS Developers look terrible to be honest.

  91. Brian says:

    That graph up there is really impressive.  Try plotting it without including IE8 and the performance difference between all the other browsers and IE9 really stands out.  600ms is a looong eyeblink!

  92. Julien Huang says:

    Ok for the same markup commitment, but for god sake, we need same font rendering commitment !

    Different font rendering between while using the same font :

    * MacOSX

    * Windows

    * Linux

    It is a nightmare for css/webdesigner, we’re juggling with IE comments and UserAgent detection via JavaScript to dynamically include OS dependant CSS files.

  93. infinte says:


    What I mean is Managed code for ADDONS, not KERNEL.

    The KERNEL needs C++ for performance, but ADDONS NEEDN’T!

  94. *Confidential* says:

    This is a little off the topic, but can you guys ensure that the vertical scroll bar is absolutely on the right of the screen? In the PP u have to carefully position the cursor to grab the bar, while in Chrome ( for eg. ), moving the mouse way to the right just gets the cursor over the scroll bar, which is quite convenient.


  95. Wurst says:

    @Dominic Pettifer: What I meant is that pixel manipulation is basically outdated and is problematic in conjunction with subpixel positioning and other ‘tricks’ like cleartype (latter for example when mirroring text).

    @Julien Huang: Font rendering should be controlled where it makes sense, e.g. anti-aliasing mode, positioning mode. ideal vs. pixel-approximated text metrics, …

    @infinte: .NET IE addons are already possible although a bad idea for several reasons (large overhead, version interoperability, memory management, …)

  96. MsftFan says:

    Nice work guys. 1 question I have is if IE9 uses Media Foundation to support H.264, does this mean it will add the H.264 Media Foundation decoder to Windows Vista as only Windows 7 has H.264 decoding as far as I know?

  97. DennieBee says:

    To start: Thanks for the IE9 platform previews. Having more interaction with your users sounds like a follow-up on the Windows7-success-story.

    Please note that I really like the improvements, especially the acceleration of graphics. Keep up that good work. For once, IE is taking lead by offering something other browsers don’t offer yet. 🙂

    What I would like to see in the next previews is some more HTML5 support. Right now some html5-features are getting community-adoption, but aren’t available in the IE9 preview yet. Among those are: canvas-tag, section elements, canvas DOM support, video element, audio element, typed form elements. Those are gaining popularity among the HTML5 examples floating around the web, but are not testable in the IE9 preview. It think it would benifit IE9 -a LOT- if you would add that support at a very early stage. It shows that the IE team is really dedicated to supporting standards (which is a very good thing) and it would enable the feedback-loop to focus on those features also.

    Keep up the good work and good luck with the development of upcoming versions.

  98. DennieBee says:

    To start: Thanks for the IE9 platform previews. Having more interaction with your users sounds like a follow-up on the Windows7-success-story.

    Please note that I really like the improvements, especially the acceleration of graphics. Keep up that good work. For once, IE is taking lead by offering something other browsers don’t offer yet. 🙂

    What I would like to see in the next previews is some more HTML5 support. Right now some html5-features are getting community-adoption, but aren’t available in the IE9 preview yet. Among those are: canvas-tag, section elements, canvas DOM support, video element, audio element, typed form elements. Those are gaining popularity among the HTML5 examples floating around the web, but are not testable in the IE9 preview. It think it would benifit IE9 -a LOT- if you would add that support at a very early stage. It shows that the IE team is really dedicated to supporting standards (which is a very good thing) and it would enable the feedback-loop to focus on those features also.

    Keep up the good work and good luck with the development of upcoming versions.

  99. Mike Dallos says:

    I am lovin’ it………works so well on my Windows 7 desktop.

    Thank you developers!

  100. toth3max says:

    Keep up the good work IE team! You’ve just passed the Nintendo DSi browser in the acid3 test (it scores 59). Although you need to catch up and add the canvas tag which is supported in pretty much every browser these days including the Nintendo DSi!

    As always I would like to remind you to share the code with the WP7 team, they will need it =)

  101. infinte says:

    Drawing images on DirectX Textures is not difficult. Get Surface and DC, then GDIPlus or DX.

    But I can’t understand WHY DON’T you provide <canvas> or something that has the same function( e.g. <stage>)? Is there any technical problem? or YOU DONT WANT TO?

  102. I really hope the eventual IE9 release will follow the Metro design concepts. The dated IE interface really needs a fresh overhaul!

  103. XP support says:

    XP use is not going away any time soon,

    like it or not. But maybe IE market share will go away if people discover they can use a

    browser like Chrome or Firefox on XP.

  104. anonymuos says:

    Just when the world was getting a standards compliant version of IE9 which is the real deal, Microsoft make us pay the Windows 7 or Vista tax. XP is well supported till April 2014 so all releases of IE within the April 2014 timeframe should support XP. Microsoft has done this in the past for by offering IE6 for Windows 2000, 98, Me and even NT 4.0. It’s because with XP people are so hooked onto the mature OS that they are deliberately using APIs that aren’t available on XP. Hardware acceleration is nice to have but not critical for browsing the web. Opera 10.53 on XP without hardware acceleration does much better than IE9 on Windows 7 and feels less laggy in Microsoft’s own IE9 tests. Upgrading the OS is the tax XP users must pay for IE9. Which is WRONG. Get rid of the OS barrier Microsoft then we will talk browser versions. It’s not my problem that Vista was not released till XP got 7 years old and Vista flopped.

  105. This all looks great. You’re making some great progress and will hopefully reach a point that’s on par with the other major browsers… and speaking as a web dev, trust me, we all want progress as much as you do!

    I have to say, though, that I shudder every time I hear the IE team mention that IE9 will "fix" the messy CSS3 "problem" other browsers create and that there will be unicorns and puppies for everyone. That ignores a key factor about CSS3 that should be close to the IE team’s heart: CSS3 isn’t a standard yet. It could change. To put it another way, *vendor prefixes prevent another IE6*.

    Let’s look at border-radius for example. It’s not a standard yet; CSS3 Backgrounds and Borders is a Candidate Recommendation. It’s almost there, but not quite. What happens if a browser implements border-radius but the  implementation changes before it’s made into a full W3C Recommendation? Well, developers would have to create a browser-specific stylesheet that reflects the outdated implementation in that individual browser. Using vendor prefixes like -webkit-border-radius and -moz-border-radius, though messy, allow developers to ensure that they can use a feature without the fear of having its implementation change later on before it is standardized.

    Personally, I don’t want to see a bunch of <!–[if IE 9]> conditionals in my sites. Seems from your posts lately that you’re using the term "standards" when it’s convenient. Take a look at what some of the other browser families have been doing lately and learn from them: they’ve been doing great work over the last few years.

  106. EricLaw [MSFT] says:

    Brian, the delta is far less than 600ms. As the demos point out, browser performance is about a lot more than JS engine benchmarks.

    Mark, you can’t set the Content-Type to XHTML via a Meta tag, you have to actually set the header. Why? Think about it… by the time IE has decided to look for a META tag, it’s already decided that your document is HTML (because otherwise, it wouldn’t bother looking for a META tag).

  107. Will says:

    <<<XP is well supported till April 2014 >>>

    No. XP is in extended support (security patches only) until 2014… it exited mainstream support (where you get new features) quite a while ago. XP is nearly 10 years old now.

  108. Jonas says:

    XP’s market share is dropping very fast indeed. Microsoft is absolutely right in dropping XP support.

  109. Aryeh Gregor says:

    @Mark Grgurev: meta tags will *not* help you.  Open up your page in Firefox or Web Inspector or whatever, and look at the HTTP headers.  You will see the header "Content-Type:text/html".  This *must* be changed to "Content-Type:application/xhtml+xml" or something like that, if you want XML processing.  In vBulletin you should be able to do this by hacking includes/functions.php.  Note that this means your site will completely stop working in IE < 9, and also if there’s any XML syntax error; you probably do *not* want to do this.

    In any event, it’s not a bug in IE9.  It’s the behavior of all other browsers too, and is standardized by the HTML5 Working Draft: <;

    @kabkab: IE8 has a CSS 2.1 implementation that’s as good as any browser’s, if not better.  What particular bugs are you thinking of when you say they haven’t gotten it right?

    @Justin Russell: See the W3C Process Document: <;  The Candidate Recommendation phase is a Call for Implementations, and browsers are expected to implement the spec at this point.  The CSS Working Group in particular has concluded that CR is the correct stage for browsers to drop vendor prefixes.  Opera already has border-radius unprefixed, and I expect Gecko and WebKit will drop it by their next major releases.  By advancing the spec to CR, the CSSWG has committed not to change it incompatibly once there are implementations.

  110. Todd B says:

    So are you going to release versions for people that *aren’t* on Windows based systems?

    If not, then why should I care about developing for IE. Hopefully IE will be 100% standards compliant (HA! Fat chance of that happening…)  so if someone comes across my site using IE, everything will look good but I’m not going to waste my time developing for something that isn’t cross-platform.

  111. PiersH says:

    @Mark Grgurev

    <script … /> is not valid XHTML. :

    "All elements other than those declared in the DTD as EMPTY must have an end tag." :

    script is NOT declared as EMPTY.

  112. @Aryeh Gregor: Thanks for the link. I was under the impression from the document that changes were still possible until the Recommendation release (see the Purpose in section 7.4.5), but it looks like you’re correct. In fact, it can’t be any clearer than section 3.3 on here:

  113. Aryeh Gregor says:

    @PiersH: <script /> is valid in XHTML.  The section you link to is non-normative (notice a bit above: "This section is informative.").  The actual conformance requirements for documents are given in section 3.1.1, and those impose no syntactic constraints beyond what XML does.  In XML, <something /> is to be treated identically to <something></something>.  If you don’t believe me, try validating this in the W3C validator:

    <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" ""&gt;

    <html xmlns=""&gt;

    <head><title></title><script type="text/javascript" src="" /></head>



    You’ll find it will validate, and browsers will render it properly if served with an XML MIME type.  The problem here is the MIME type — the syntax is correct.

    @Justin Russell: Changes are indeed still possible in CR.  CSS 2.1 is an example of a spec that reached Candidate Recommendation once before being demoted to Working Draft.  But the CSSWG is not going to make significant incompatible changes to a feature after it’s been implemented, unless there’s a really dire need.  Not because of W3C policy, just that it would waste implementers’ time and worsen compatibility for authors.  To avoid getting stuck too early, it asks vendors to use prefixes before CR, so it still has flexibility to change things if design flaws are pointed out.  But it won’t change things once they’re implemented unless it’s really really necessary.

  114. Tom says:

    I dont understand why people are complaining about a 10 year old OS! Come on people, move on already! Windows 7 + IE9 will rock!

  115. JoergW says:


    will you support SMIL or not? Which ACID3 value is defined in the product definition of IE9?

    Best Regards


  116. James says:

    @Tom "I dont understand why people are complaining about a 10 year old OS! Come on people, move on already! Windows 7 + IE9 will rock!"

    The features in IE9 are certainly welcome, however the likelihood is that the features that IE8 lack will have to be coded around for years, just like IE6 is still done today at many places when the usage is high enough it can not be ignored. Your response might be more shared if it has not been sold in ten years, but XP has still been a popular option in the last couple years, and their are a lot of people not prone to update any OS even if it were free over the life cycle of their machines.

    This either aversion to updates or ignoring is mirrored in not updating to the newest browser version they commonly use. This is why I which the default option for IE was install updates in the background (independent of windows updates) and if the user (or corporation) chooses then either be prompted or never alert/auto-install.

  117. DanielHendrycks says:

    Any canvas support?

  118. serpent says:

    Hey guys…good job! now, will you release ie9 for windows 3.1? if you don’t, I am going to cry. :'(

  119. Co len IE9!!! Co len Microsoft!!! Hay danh bai Google Chrome va Mozilla Firefox 🙂

    Comment tren browser Chrome 5 beta 😀

  120. says:

    Отличный браузер IE9. Мне понравился! 🙂

    Буду тестировать…

  121. Chrome says:

    Chrome 6.0 still is faster than IE9 PP2.

    AppleWebKit/533.8 (KHTML, like Gecko) Chrome/6.0.399.0

  122. CANVAS says:

    "One other topic of conversation from the panel was IE9 and Canvas. I asked the crowd who would like to see canvas supported in IE9 (didn’t show up in yesterdays preview). 80% of the crowd put their hands up, and I am sure the other 20% just didn’t get them up in time 😉 Giorgio of Microsoft himself put his hand up, which was nice. The strange comment from him was that the canvas spec is "big" which was refuted by Brendan. Microsoft has implemented SVG support which is massive in comparison, and I am sure they could wire together canvas API calls to the underlying 2d engine ;)"

    Microsoft still doesn’t support or respect web standards. Same markup? I believe it when I actually see Canvas support.

  123. Matt says:

    No, that statement was not "refuted" by anyone (you should look up the word "refuted").

    Brendan disagreed, but that doesn’t make him correct, particularly since the term "big" is entirely subjective– it’s at least 30 printed pages, not counting the references, and "big" can refer to more than just the literal size of the text. Microsoft is doing the editing for that spec, so they know exactly how big it is.

    The SVG spec is *huge*.

  124. Badger says:

    @toth3max  The Nintendo DS/DSi use Opera. So while your comparison was cute, it’s not that impressive. Opera is the flipside of standards support:  too strict and not realistic (so strict that many real web pages don’t work right).

    @*Confidential*  The platform preview is not at all the UI for IE9. So while it may have some bugs with the scrollbar as you mention, it’s not likely those will be present in the actual release of IE9 (or the beta). They’ll have completely different UI code.

  125. sy says:

    Please add the following:

    1) text-shadow. Blurred shadows are not possible with DropShadow filter, and IIRC filters are being deprecated.

    (On the contrary, box-shadow is not that important since it’s easy to create with images.)

    2) CSS Transforms, e.g. "transform: rotate(90deg)". This is necessary to advance web designs and will be a great showcase for IE9 GPU acceleration.

    People should not be using SVG just to rotate a single picture or a Post-It-style block of text.

    Imagine a web gallery styled via a user-modifiable stylesheet. Then, a "casual" look could be achieved as easily as:

    :nth-child(3n) { transform: rotate(2deg); }

    :nth-child(3n+1) { transform: rotate(-3.5deg); }

    :nth-child(3n+2) { transform: rotate(3deg); }

    Or, see for transform usage in a weblog theme.

    As for the SVG/canvas debate, I’d rather see a rich and compatible SVG implementation first (instead of two half-baked implementations). Actually, SVG is much more HTML-style. As opposed to canvas, it is zoomable (and therefore adaptable to coming-soon high-resolution displays), and it has a DOM, allowing introspection, mutation and attaching interaction events.

    Personally, I’m totally OK to see canvas(‘2d’) in IE10, together with other programmatic technologies such as canvas(‘3d’)/WebGL and yet-to-come audio buffer APIs and camera/mic APIs.

  126. Aryeh Gregor says:

    @Matt: The Canvas 2D Context spec <; is edited by Ian Hickson, like (the rest of) HTML5.  Maybe you’re thinking of something else.  It’s more than thirty pages, but tiny compared to SVG, which Microsoft implemented (from scratch?) for IE9.  I don’t personally know anything about 2D graphics, but Robert O’Callahan of Mozilla has said (from experience, I take it) that "implementing canvas on top of Direct 2D is really quite easy".<;

  127. Matt says:

    Editing != Editor.  See

    "really quite easy" != "not big", and I’m not sure Mozilla has a lot of credibility when talking about IE.

  128. Wurst says:

    @JoergW: Some of the SMIL features can be found already in HTML+TIME

    @Badger: Opera being too strict? I remembered Opera  as the browser implementing many IE quirks and MS proprietary extensions like document.all and such…

  129. Ahmed says:

    greetings IE team

    guys i have some 1 feature request

    would u make a stripped down version of the final IE9 product to upgrade windowsXP ie6-8 without the d2d support (GDI) this is better than many people switching to chromeFrame ?!

  130. MacGyver says:

    Very nice progress IE Team!

    Any chance on the IE9 Preview #3 also being released in 64bit flavour ?

    What are your plans on moving plugins to their own boxed process so that they can’t bring down the entire browser anymore?

    Likewise will any enhancements be made to plugins not affecting general browser performance in any way. An example with IE8 is that having the Java plugin enabled, causes new tabs to be opened a lot slower (perceptably).

  131. Doug M says:

    You know this would be NICE if Microsoft FINALLY puts out a CSS3 compliant product.  I was so fed up with the bugs in IE8 that I went back to IE7 and refused to sue IE8 no matter how many times Windows Update suggested I download it.

    Microsoft currently holds a horrible reputation with web designers because of the difficulty in W3C compliant web design.

    I would be pleasantly suprised if IE9 actually meets the standard, but quite honestly, I am in the "I will believe it when I see it" mode.  Excuses I saw in the various trade shows I went to and in the press releases did not impress me at all.  Seemed too much like Microsoft was bucking responsibility for meeting consumer demand for a W3C compliant product.

  132. Gopal says:

    XP users (me included) stop complaining. So what if IE9 is not available.  You should try out google chrome frame which runs within IE. This is continuously updated and works well.  Few issues relating to printing yet to be sorted out but otherwise is good.  This will any day beat IE9.

  133. Crescens2k says:

    Doug M:

    What would you class as a CSS3 compliant product when the majority of CSS3 is still in the working draft stage.

    Whats more, CSS2.1 and CSS3 selectors are being treated as high priority where the rest are normal priority or lower.

    So right now there is no way to say what a CSS3 compliant product is since there is no standard.

    CSS2.1 and CSS3 Selectors would be seen as more of a worthwhile risk to add since CSS3 selectors is at proposed recommendation and is moving onto recommendation. CSS2.1 is built on top of CSS2 (which is already a reccomendation) and the CSS2.1 document is moving from candidate recommendation to proposed recommendation. Or would you be happy if they implemented a working draft to find at it gets changed in a big way? I’m sure if Microsoft did that, there would be even more of an outcry because of bad compliance, or Microsoft doing their own thing again or something similar.

    Come back and say to implement CSS3 when more of the documents are finalised so there is much less of a risk that it will get changed.

  134. MacGyver says:

    Gopal, did you even try out the tests on the ietestdrive page in Chrome? IE9 is already beating Chrome as it is. Why can’t you people give credit where credit is due. Microsoft is giving IE the Windows 7 treatment and it’s going to be good for everyone; from consumers, website developers, IE devs to Microsoft as a company.

    PS: bye bye XP, you won’t be missed.

  135. Rob Crowther says:

    @Aryeh Gregor: SVG is based, in part, on VML and IE has had VML support since version 5.0, so they may have been able to use that as a starting point.

  136. Neuro says:

    People, stop saying "xp users, get w7". That’s not an issue here. Most of people reading this blog probably have vista or win7 anyway. The issue are the ‘normal’ people. Our page has now 55% of xp users, and that number has almost not changed in a year. If ie9 isn’t available for xp, then I am looking towards years of coding for ie8 (which was outdated before it even came out). I don’t know when ie9 will be out but I doubt the situation will change substantially in a year. Please, don’t let ie8 become the new ie6

  137. wai says:

    beside performance, would IE team consider to add some functionality benefit to the end user?

    – native spell check, rather than 3rd party installation

    – easier way to write extension, com/.net based still cannot archive the rapid growth like firefox and chrome do

  138. madriss says:

    Will the debug menu to allow you to render as IE5-IE8 be kept in the final version? This would make an amazing developer tool. How come IE6 is not there? If we had IE6,7 & 8 rendering modes in-built and could use developer tools with them it would be incredibly useful!!

  139. EricLaw [MSFT] says:

    Madriss: The IE Developer toolbar (available by pushing F12) allows you to select a document mode. IE6 isn’t there (or in the PPB) because the IE6 rendering code was replaced by IE7’s engine. When we built IE8, we recognized the utility of keeping older modes available for compatibility reasons. Hence, in IE9, you have Quirks (~IE5), IE7 Standards, IE8 Standards, and now IE9 standards modes available.

  140. Wurst says:

    @madriss: You might be interested in the tool Microsoft SuperPreview that is part of the Expression Package but also distributed freely at

  141. wos says:

    I wish Microsoft would make a browser to compete with Google Chrome.  The reason that IE continues to lose market share is not just because Chrome is technologically superior, it is because, too, that people associate IE with slowness, security issues, bugs, non-standardness, etc.  As long as Microsoft continues to stay on the IE bandwagon, IE will continue to lose market share.

    I say, it is time that Microsoft make an entirely new browser with an entirely new name that really gives Chrome a run for its money.   But MS won’t do that; they will continue to make IE, and it will continue to lose market share until Chrome rules. Another market lost to MS stupidity.  Next up: the OS market.

  142. Blah blah blah says:

    Yes, Microsoft needs to give Chrome a run for their money. I mean, Chrome has 1/10th the marketshare of IE, so I’m sure the IE team feels terrible to be losing so bad in the marketplace.

    Oh, wait, that doesn’t make any sense, does it?

  143. kudraw says:

    A little off-topic about a feature request:

    Why don’t put an option to isolating IE in a Secure Desktop (on user request)?

    It can be useful for delicate operations and transation.

    Thank you, bye!

  144. madriss says:

    Thanks for the quick reply Eric, I guess what I was suggesting is if "IE6 Standards" was added there it would be really useful and we could do all web development from within just IE itself then whilst supporting IE6. At the moment I develop for both IE8 and IE7 using IE8, and find the in-built developer tools incredibly helpful for this, but IE6 is much more difficult to develop for without being able to use the developer tools to alter things on the fly and figure out what’s causing the various issues that inevitably come up.

    Thanks for the heads up on SuperPreview, Wurst. I’ve downloaded it and it’s certainly a useful tool. Think I looked at it once before but did not notice the Dom feature, I’ll probably start using that now. If we could alter CSS on the fly as with IE Developer Tools though that would make my job a hell of a lot easier!

  145. Lance says:

    Sorry – Off-topic, but anyways…

    In the dev tools, please put a JS DOM browser like firebug has so you can inspect JS objects from the window object down. This is very handy for debugging code as I often store references to objects in global vars/objects (ie references to YUI elements).

    Also, for the new network monitor, please add a right-click option to open request in new tab/window. If you are debugging a problem with a XHR call, sometimes it’s nice to just open the request in a new window and hit F5 until it’s fixed.. Must work with POST too.. Fiddler doesn’t let you do this (in an IE tab anyway).

    Lastly, for the user interface, please rip-off chrome’s feature of dragging tabs into new windows and vice-versa.. this is very handy for development.

  146. @Lance: "Fiddler doesn’t let you do this (in an IE tab anyway)."

    Sure it does, just tear-off the Request Builder tab (click the button on the Options tab) and set your Response Inspector to the WebView tab.

  147. Bruce Leggett says:

    Why is it still slower than old competitive browsers? The average end-users really emphasize speed. That’s where they spend there money; and, that is where businesses make their money. Customers always ask, "Is there anything you can do to make my computer faster?" SPEED! SPEED! SPEED! Speed is three times as important than the other "new" features.

  148. Phil says:

    Bruce, are you a troll, or did you simply not try the demos? The preview is *waaay* faster than Firefox and Chrome on things users care about… users don’t sit around and run artificial benchmarks like sunspider…. boring!

  149. Phil (an other one) says:

    Doesn’t matter how fast your browser is if you’re on a laggy OS (looking at you Vista) or your internet bandwidth is being hogged by the kid next door downloading movies or your family facebooking and twittering and that other nonsense. The browser is the last link in a long chain of speed-affected components – and seriously, does IE 9 being half a second slower or faster on something really make that much of a difference?

  150. I.B.Worried says:

    Please check this out guys.

    If you want the tech community to take your commitment to HTML5 and CSS3 seriously, you have to improve on graphs like these.

  151. I just downloaded the new preview and tried viewing my site which uses CSS3 extensively and IE9 fails miserably.

    Firefox,Chrome & Safari all display my site pretty close to the same with each not showing something correctly. Only Chrome/Safari show everything as it should. IE9 is missing slightly rotated pics and text, complex 5 layer shadows on text, shadows and rounded corners on containers, even the HR messes up. Most links don’t do the changes they should and most pics above links are not clickable.

    Check out my main page at with Chrome/Safari then look at it with IE9.

    I didn’t even get around to testing local databases which already work great in Chrome.

    I used to be an IE promoter since IE3beta because you were always leading… now you have fallen way behind… Please start supporting ALL the standards not just the ones you want.

    I will keep trying IE9 till it releases but it is not looking good from what I see right now.  

  152. I just noted that my demo area of my website(uses classic ASP written for IE4+) doesn’t work unless I set the browser to IE5 yet I can view the demo area of my site with no problem using IE7 or IE8. Not sure what causes this but site was very advanced for it’s time.

  153. Sarah says:

    You’ve done impressive work with CSS3 to make a site that’s ugly in all browsers. the 9 preview supports rounded corners; you’re doing something wrong. it doesn’t support shadows. "IE5" == quirks. If you don’t know that, and don’t know that the fact that you used "ASP" is entirely irrelevant, I kinda doubt you’re really a professional web developer.

Comments are closed.

Skip to main content