Tuesday, September 02, 2008

Google Chrome: A first little test with highly complex AJAX/Javascript web applications

Google Chrome - what is it's real performance for highly AJAX-based web applications?

Some five hours ago Google introduced Google Chrome and put it on their site for download. Google Chrome is a stand-alone web browser and therefore has to compete with browsers like Internet Explorer, Firefox, Opera and Safari. Here at first a core Javascript test (as stated in a previous Javascript-post):

Chrome 0.2FF RC1IE 7.0.5730.13Opera 9.27Safari 3.1.1
Time in milliseconds the browsers needed in this performance test

But Google Chromes main goal is (according to Google) an other goal than that of other browsers: Highly AJAXed Javascript "web applications" shall run faster and more safe with Google Chrome, thus enabling a new era of web based applications like Google Maps, Google Mail or others. Time to look at speed and compatibility issues with more complex applications, even though the browser is stated as a "Google beta". Again.

YAWB or not YAWB, that is the question.

Yet Another Web Browser? Or: Why is Google in the web browser market? Because that is Googles business, the idea fits. Why not presenting the browser as a gift and earning money with the applications they run?

Internas and testing environment

Time to do a kind of real life test using an internal application that makes heavy use of Javascript and AJAX. At the moment Google Chrome is too much of an early bird to do this excessively, but we can have a look under the hood and run a fairly complex application on it.

The new features include
  • A new Javascript Engine called V8 is in use.
  • Better performance through compiled Javascript, it is compiled into machine code and not necessarily always interpreted.
  • The new Javascript VM has a better garbage collection included to preserve memory.
  • Browser tabs or new browser windows run in dedicated threads.
The testing application has some thousands of lines of self-written Javascript-code and relies on Prototype and other frameworks. It (re)opens different "windows" as DOM-elements and is able to switch through them via ctrl-tab and shift-ctrl-tab. Data is drawn from the background via AJAX, the usability allows drag-and-drop of single (or multiple) elements from one window into others. The application "runs" on IE, Firefox and Safari. Pretty complex stuff for a first test.

Compatibility will be the key...

What use is a new browser if the sites you want to visit will not work with it? Right, none. So the browser has to be as compatible as possible. I was fairly surprised and impressed as I saw the application running smooth and dandy. There was no itch regarding the Javascript-compatibility, every function I tested simply worked. The whole Javascript-code ran without the need to apply a single patch or something similar from our side. You know the moments where you hope "ah, come on, it will work out somehow. I believe, I strongly believe..." and you run it and it works out? That was one of that rare moments. Impressive.

... and performance will be the door-opener.

So it seems to be pretty compatible, right, but the second side is, is it faster, more useable than other browsers? The rendering engine seems to be partly from the webkit.org - project, rendering was pretty fast and compatible (to say the pages looked like in other browsers). From the look and feel of the thing I would say Google Chrome is at least one of the fastest rendering browsers if not the fastest, anyway.

And the Javascript-performance was astonishing (see here for numbers of a normal Javascript test suite). The application, running pretty fast in Firefox 3 and Safari 3.1 got a little extra speed-kick. That way it really feels like a real application on the computer rather than rendered web results. The clear separation of tabs and windows into threads does its additional work: Because no tab has to wait until an other tab releases the processor you have the feeling that it works faster or smarter.

So is it worth the hassle?

I think it is worth giving it a try. I do not think that I will use it yet as a browser-user, because I need my Firefox-plugins or IE-updating. But I will definitely test it with/for Javascript and AJAX-driven pages and our own developments. The gain in there is definitely worth the hassle.

Labels: , , , , , ,