Google is launching their own open source browser tomorrow

There is no end to the innovative spirit that is Google. That is one amazing company. I was delighted to learn of the new browser, and just might try the beta!

This is all "blue potatoes" to me, but perhaps useful to some Mac users.. and maybe others.

http://www.appleinsider.com/articles/08/09/01/google_planning_new_chrome_browser_based_on_webkit.html

(illustrations left out)

Google planning new Chrome browser based on WebKit

By Prince McLean
Published: 03:45 PM EST
A Google blogger has published an illustrated document outlining Google Chrome, a new web browser project based on Apple's WebKit open source rendering engine.


The blogger says Google mailed out a description of the new project's features in the form of a 38-page comic book, illustrated by artist Scott McCloud of Understanding Comics and distributed under a Creative Commons license.

The comic book portrays Google engineers highlighting stability, speed, security, and simplicity as key aspects of the project, which will serve not only as a new browser but a foundation for a new generation of JavaScript thick 'web client server' apps following the same model as Apple's MobileMe apps built upon the SproutCore framework, as well as the HTML 5 advancements going into Safari 4 to make it suitable for a new class of sophisticated web applications.

Most of Google's existing products are web-base applications. A Google browser has been long rumored, in part to keep Microsoft from controlling all access to the browser platform across Windows PCs. Google has been financing Mozilla Firefox development with the same intention.

While the www.google.com/chrome URL is referenced in the report as not being valid yet, Google has already released developer tools for Gears, an effort to improve web applications by providing features missing in today's browsers, including a database engine and support for local storage and offline applications. Chrome is simply, well, the chrome on the gears.

Chrome and Webkit

The report says the new Chrome browser will combine Apple's Webkit, Google's Gears, and a new JavaScript virtual machine called V8, intended to dramatically speed up the performance of JavaScript code, particularly the code used in building significant web apps. The document outlines that Google's Chrome team selected Webkit for the same reason Android developers did: it's fast, simple, uses memory efficiently, and "it was easy for new browser developers to learn to make the code base work."

However, rather than speeding up JavaScript parsing as Apple and Mozilla are doing in their efforts to build faster JavaScript interpreters, Google's V8 compiles JavaScript into native source code. It also allows for better garbage collection of expired memory and dynamic optimization based on hidden class transitions. The company will offer V8 as an open source component that other browsers can adopt as well.

This puts the Chrome project on the same page as Apple's forthcoming Safari 4, which similarly intends to speed JavaScript (although in different ways) and push features of HTML 5 to give web apps standardized access to database tools, local storage options, and the ability to work offline when the network isn't present. Conversely, Google's use of the Webkit rendering engine will also help popularize Apple's efforts to spread its open new CSS features for adding effects such as reflections, gradients and new masking capabilities.

Chrome user interface

By delivering its own browser front end, Google can focus on differentiated user interface features. The illustrated report depicts a browser with tabs on top of the window rather than inline, a violation of both Mac and Windows user interface guidelines. Apple, Adobe, and Microsoft have all taken similar exceptions to established guidelines in order to make their apps stand out. In Google's case, the top tabs are intended to make it easier and more obvious how to detach tabs to create freestanding windows.




A separate example cited in the report is a browser address bar with auto completion called "omnibox,' designed to respond to natural language search words that match previously visited pages, rather than force the user to enter semantically correct URLs.

Another user interface idea presented in the illustrated guide is Chrome's default page, which shows the user thumbnails of sites they have previously visited or bookmarked and recent searches (below), a feature similar to one in the Opera browser.




Like Safari's Private Browsing introduced in 2005's Tiger and Internet Explorer 8's new InPrivate mode, Google's Chrome will support 'incognito' windows and tabs where the browser won't save any history, cookies, or leave other private tracks. And similar to an upcoming feature in Safari 4, Chrome will be able to launch web applications as a freestanding desktop app lacking the usual browser address and tool bars.

Targeting responsiveness and bloat

In addition to visual enhancements of the user interface, Google is also attacking the inherently single-threaded model of browsers, which results in delays while the browser waits for a JavaScript to finish executing. Rather than making the browser multithreaded, the report depicts the new browser using a multi-process design that runs each browser tab as its own process.




That design would chow down more memory initially, but provide much better response time and independence between web pages and applications loaded at once. Closing a tab would also immediately release its memory allocation, rather than resulting in a fragmented allocation that affects the entire browser.

The document cites Google engineer Brett Wilson as explaining, "so as you browse, we're creating and destroying processes all the time. If there's a crazy memory leak it won't affect you for that long because you'll probably close the tab at some point and get that memory back."

Chrome and web plugins

Each tab is also sandboxed for security, although any web plugins, such as Adobe Flash or Microsoft Silverlight, will not respect this security model by design, because the existing web plugin model automatically gives them privileges on the same level or higher than the browser.

The document notes that "with some small changes on the part of plugin makers, we can get them to run at a lower privilege which would be much much safer." It also notes that Chrome will present each plugin in the browser's task manager so users are aware of which plugins are hogging available RAM, "placing blame where blame belongs," a shot clearly fired toward fat third-party plugins.

While the document carefully avoided any mention of Flash, it did depict the challenge of isolating video within YouTube, a task that is built on top of the Flash player plugin.




Popups, malware, and bugs

Rather than allowing JavaScript to popup ads from any web page in the background or in the user's face, Chrome will attach JavaScript notices to the browser window and tab from which they originated.

The new browser will also get updated info on malware sites to help rapidly warn users of phishing attacks that have been discovered (phishing schemes usually get taken down with a day or a few days). Google will also be exposing this system as a public API other browsers can use.

In order to track down bugs and problems with various web sites, the document outlined Google's ability to automate testing of new builds against tens of thousands of websites within minutes. It also notes that Google's own Page Rank information will be used to direct testing toward the popular sites people actually use regularly.

Wide open

By offering Chrome, Gears, V8 and other components as open source, Google hopes to bring all web browsers up to modern standards capable of running the kinds of web apps the company is delivering now and laying a foundation for a future of even more sophisticated apps that run on the web platform using interoperable standards anyone can implement.

Google's Chrome will give a huge push behind the efforts of Firefox, Opera, and Safari to create a more open, interoperable web enabled to run a sophisticated new generation of web client server apps.

The entire comic book is available at Google on Google Chrome comic book

Mistake. All this could be is something which interacts more strongly with Google's search engine, which is simply not that well conceived or good.

That's not all it can be, and they don't need their own browser for that. They already pay other browsers like Firefox and Opera for the default search boxes on those browsers (e.g. over $50 million to Mozilla last year).

The real innovation line may be a bit too technical for you to grasp, but is hinted at strongly in the name of the browser. If you look at the differences between Mozilla Prism and Mozilla Firefox you'll see what directions the browser will go soon.

Now that doesn't mean Google is going to get it right or anything, but you don't seem to understand the next generation of browsers at all. It's about persistent local storage and in breaking the web apps out of the traditional browser box.

Sure, they want better integration with their own products, but they'll offer the protocols up for everyone else to use as well and they aren't the only ones working on the next generation of browsers.

They made the comic public and you can find out more here:


http://books.google.com/books?id=8UsqHohwwVYC&printsec=frontcover#PPA6,M1

Wow, this could be big. Wish I heard about this sooner, or had more time to read up about it tonight. I skimmed through the comic, and saw some interesting stuff. Separate processes for each tab is awesome. I read they took stuff from Mozilla and Webkit.. It will be interesting to see how much. How standards compliant are they?

Exciting stuff

Thanks osso, this was a much preferred format then a comic for me to read about it. So it will be webkit rendering. The v8 engine sounds interesting. I wonder if it compiles all javascript to native code, or simply selected web apps (google own, to start?). I wonder if there will be an initial performance hit while it compiles the javascript, and how caching of javscript will work.

Firefox's recently released TraceMonkey JavaScript engine for Firefox 3.1 introduced Just in Time compiling for JavaScript and you can see benchmarks versus their previous SpiderMonkey engine here:

http://weblogs.mozillazine.org/roadmap/archives/2008/08/tracemonkey_javascript_lightsp.html

I haven't played with it yet, but it does sound like they nailed it. SquirrelFish is another recent JS engine to check out that has reportedly improved WebKit's benchmarks.

Edit: this comic format is irritating, but I've read enough about it now that it seems like it's going to take both the VM and the JIT compiling approaches in addition to other interesting things like their hidden class transitions.

Especially interesting is their testing process. They have a "chrome crawler" which tests the site automatically on millions(i forget the actual number) of the most visited sites on the internet. Cool stuff.

What would be really interesting is when they start using their parsing engine for their search indexing. Instead of reading the document markup and trying to catch hidden text and cloaking algorithmically they could use the parsed output and see what the user sees in their spider.

That would be a very interesting change in search engines and I think they have it in mind.

First to respond from Chrome! I'm using it now (this is the first page I went to). Don't have much of an opinion yet, but the install was easy, and interface is intuitive. Status bar only seems to be there when you need it, but not all the time. I liked the address bar (whatever they call it). As it has google suggest built in. As I started to type able2know, it knew I wanted able2know.org, and suggested it.

My first impressions:

- It has a lot less chrome than other browser defaults.
- It renders quite well on the sites I have tested.
- It has a lot of nice UI touches for BETA software, taking a lot of the best from other browsers.

But whether or not I'd use it more comes down to whether or not it can generate an ecosystem of plugins to rival Firefox's.

Interestingly, they are advertising it on their home page (you may not always see it) and aren't browser sniffing well, because using their own browser I see an ad to download it.

I agree about the plugins. Firebug is a big one for me. However, I did find that Google is already halfway there. Try right clicking on an element, and selecting 'Inspect Element'.

You are presented with a DOM view of the page, and selecting an element will give you all CSS properties for that element. Click the wierd blue icon in the bottom left to bring up a console, where you can run javascript. No word yet if firebugs' console.log works here.

Lastly, clicking on the Resources tab, then reloading the page, will show you everything that was downloaded, and how long it took. It will even show you if certain items have any warnings or errors.

Okay, I hope plugins are made for this. The browser rocks so far!

History search works great. After posting here, I opened a new tab, and saw a search bar for your history. I entered "Nick Ashley" and it pulled up all pages that had my name in them, including the pages I was just on. Firefox only searches through domain name, title, and any tags you've given it. This seems to search through the actual pages itself. I could see this being very handy when looking for something. (ex: I remember it was about X, but thats it...)

I think that it's great though I haven't used it longer than 12 hours now.
Still a lot to discover ... (This search function, Nick, really astonished me - seems to search ALL the sites you've visited.))

Whats there is nice, but you cannot customize what is there. I much prefer scrolling tabs to squishing tabs when you have more then 10 or 15 open...

There are a couple of "Hidden" about pages I've found:
about:stats
about:memory

But 'memory' isn't

Have they done any integration with their own suite of apps? I find it funny that it tries to import bookmarks from IE/Firefox, when my bookmarks are in Google Bookmarks. Is there anyway to get Chrome to use Google bookmarks the same way IE and FF do?

The history stuff isn't so amazing if you're used to using google desktop search. How blase I get. Bring back Mosaic!!!!

There's also about: (for basic version info)

But I can't find an about:config equivalent (tried conf, config, options, configuration...)