How is this different from Microsoft's ActiveX technology? Security wise, ActiveX relies on trust, i.e. it's up to the user to give the ActiveX control permission to run. Native Client on the other hand relies on automated sandboxing and code analysis to make sure that the application is safe to run. Native Client also takes a more open approach, by open sourcing the technology. Native Client is portable across the x86 architecture, which includes Windows, Linux and Mac machines, but it potentially leaves important mobile platforms in the cold.

Initially, the main target for the project is resource intensive applications, like physical simulations and visualizations. The image above depicts the Mandelbrot set, as generated by a port of the fractal viewer application XaoS to Native Client. The 3D game Quake has also been successfully ported to Native Client. Generally it has proven easy to port existing C/C++ applications to run in the trusted environment, which is really promising.

If the launch of the Chrome browser was Google's first step towards making the OS desktop obsolete, Native client is yet another step in that direction. Soon we might be able to do all computer tasks, even resource intensive ones, inside the browser window.

Related articles by Zemanta

• Google Native Client - A Detailed Discussion

• There was some discussion this week about which products and technologies are really threatened by Google Chrome. I agree with those who argue that other rich Internet application frameworks (RIAs) are the technologies at greatest risk. These include the proprietary Microsoft Silverlight and Adobe Air (with Flash). The great promise of the Chrome browser is that it pushes for the open technologies JavaScript and HTML. Chrome comes with a fast JavaScript engine, and with improvements in the JavaScript language itself, as envisioned by the recent ECMAScript Harmony agreement, JavaScript could become a real challenge to the programming languages used in Silverlight (C#) and Flash (ActionScript). ActionScript and JavaScript have the same roots in ECMAScript, but ActionsScript requires a proprietary runtime component (Air or Flash player) to run in the browser, whereas support for JavaScript is built-in into most browsers. Silverlight also requires a proprietary runtime component.

  JavaScript can not alone pose a threat to Silverlight or Flash, an enhanced HTML is required, with elements from the emerging HTML 5 standard, such as the canvas element, for drawing to the screen, and the video element, for displaying video. The canvas element is currently supported by WebKit, the HTML rendering component used in Chrome, and by Gecko, the one used in Firefox, but not natively in IE, though there are workarounds. The latest Firefox 3.1 alpha 2 release includes support for the video element.

  Another advantage with Chrome is that it comes included with the browser extension Gears, which is a JavaScript framework that equips the browser with additional capabilities like offline access for supported sites.
  Tags: Chrome, JavaScript

• Last week I posed a question regarding the relative performance of Chrome's JavaScript engine V8 compared to the latest Firefox engine TraceMonkey. Now John Resig has given a balanced answer, and it turns out that V8 and TraceMonkey are quite comparable. V8 is faster in some tests, in particular those involving recursion, while TraceMonkey is faster in some other tests. For tests including both JavaScript and DOM manipulation, WebKit based browsers like Safari and Chrome are somewhat ahead of TraceMonkey and Firefox 3.0.1. IE is generally lagging behind.
  Tags: Chrome, Firefox
• A proof that Chrome is relatively compliant with emerging Web standards is that it performs well on the Acid3 test, with a score of about 78 out of 100.
  Tags: Chrome, web standards
• Finally, Ted Dziuba challenges my comparison of Chrome to an operating system last week, though he probably didn't read my article, instead he mainly goes after a post by Michael Arrington who labels Chrome "a full on desktop operating system that will compete head on with Windows." Though that expression might be somewhat over the top, I enjoy the vision in the post that eventually the need for a stand-alone desktop operating system will disappear, and that basic OS features might as well be integrated into the browser. A possible solution could be based on a stripped-down version of the Linux OS combined with Google Chrome.
  Tags: Chrome, vision

Anyhow, when I first tried Google Chrome I was bothered by occasional periods of high disk and CPU activity, which could last for minutes, or at least it felt like that. Since my hard disk is very loud when reading and writing, it really was bothering. Also the CPU activity was high at those occasions, which raised the CPU temperature, causing the CPU fan to spin, contributing significantly to the noise. This was a real showstopper for me, which felt a bit sad, since I had otherwise enjoyed the experience with Chrome.

Fortunately, after poking around the limited options settings in Chrome, I found a solution: Disable the "Enable phishing and malware protection" option, which is found under the Security section of the "Under the Hood" tab. This solved the problem completely, and Chrome is now very light on CPU and Disk usage. Total CPU usage for a days worth of work only amounts to a few minutes, which is very satisfying.

Of course, disabling a security feature is at your own risk. But in my case it was a matter of being able to use Google Chrome at all.

There's a Lifehacker post with a bunch of other tips on using and tweaking Chrome. Look here for a full list of keyboard shortcuts.

• Each tab runs its own process, not just its own thread as expected. This means a great deal of separation between the tabs, if one behaves badly it will not affect the other tabs. A drawback of this approach is greater resource requirements up-front for each tab, in terms of memory allocation etc., but it also means that it's more easy to do clean-up when a tab is closed, possible memory leaks e.g. will not remain.
• To handle the individual processes there is a process manager, which in part resembles an actual operating system. Details about how the process manager works and cooperates with the actual OS are not covered in the book, but there is a task manager, where you can look at the processes and the resources they are using, and terminate ill-behaving processes. Much like the Windows Task Manager.
• The web page rendering engine is based on the open source engine WebKit, which powers the Safari browser and the one in Android.
• To boost JavaScript performance Google hired a special team from Denmark, V8, to create a new effective Virtual Machine for JavaScript. Features include "hidden class transitions", which identify class like structures and perform dynamic optimizations based on that, and dynamic code generation which generates just-in-time compiled code for faster execution, probably similar to the JIT compilation in Java and .NET. The garbage collection of unused objects has also been improved, both in terms of speed and efficiency.

This news is currently all over Techmeme.