The development direction of Microsoft IE9 is very clear, supporting HTML5 and supporting standards. At this week's MIX10, IE project manager Dean Hachamovitch said so. In the development of IE9, Microsoft began to focus on real-world needs. They monitored the top 7,000 websites in the world in real time, analyzed every JavaScript and DOM API they used, and provided comprehensive support for them.
Behind these 7,000 use cases, many are related to HTML5. Microsoft did not disclose further details, but considering the weight of HTML video and SVG animation, it is obvious that they will be the core of future HTML5 websites. However, the focus on HTML5 does not mean that Microsoft will spend a lot of effort on tests such as SunSpider and Acid3. With the development of IE9, although their Acid3 score will continue to improve, this is not what IE9 values. Acid3 is a mechanical test, a non-systematic test. You can do very well in browser performance but fail the test, or you can do mediocre indicators but get full marks on Acid3. Although the Acid3 test is useful, But it's not exhaustive, especially since hundreds of HTML5 features are not included in the test.
In fact, there are many levels of support for standards. CSS rounded corners seems to illustrate this problem. Both IE9 preview and Webkit support CSS3 rounded corners. Firefox's Gecko engine also supports rounded corners through an extension. Rounded corners are a dream feature for web developers. However, the rounded corners function does not perform consistently on different engines:
The picture below is what the designer wants to see
This is what developers don’t want to see
The undesirable appearance above is the actual performance of some browsers that support CSS3 rounded corners. There are two possibilities. Either the browser is wrong, or the standard is not strict enough. However, Acid3 cannot test such a problem. Yes, this problem requires a systematic test to solve.
When designing this type of test, you can find problems in some standards and errors in browser implementation. Acid3 cannot test a browser's support for HTML5, or even tell you which HTML5 functions the browser does not have. Yes, this type of issue requires a more comprehensive test to resolve. It is for this reason that Microsoft has continued the work they have done with IE8. In IE8, they submitted a large number of CSS2.1 tests to W3C. Systematic testing is the only method that can ensure that real-world needs are met. Therefore, in the development of IE9, Microsoft designed a new testing method. They have submitted the first batch of tests to W3C. Microsoft does not want IE9 to compete with other browsers for scores.
Although there are some performance differences between browsers, the gap between them is not as big as IE8. In any case, all browsers should strive to make the same code perform the same.
In addition, SunSpider's JavaScript score is not the target of IE9. The SunSpider test can reflect the performance of the browser's JavaScript engine in many aspects, but just like real web pages are not designed according to the Acid3 test, real Web applications are not designed according to SunSpider. As written, real Web programs need a lot of optimization to allow web pages to load quickly and to allow complex operations to be performed quickly in the background. SunSpider cannot test these things. SunSpider does not represent the real world, and developers should develop for the real world, not a test platform.
Microsoft hopes that their support for HTML5 will be stable and robust, which means that IE9's support for HTML5 will not be comprehensive. There are still many variables in HTML5. In addition, some HTML5 functions, although reliable, are not among Microsoft's 7,000 use cases. The final IE9 may take a different path from other browsers. If IE9 has any further discrepancies in standards support, it will be a big problem for Microsoft. However, IE9 is currently heading in the right direction.