Weekly Edition
Return to the Development page
| |||||||||||||
The Google Chrome comic book
Google Chrome, being the new Webkit-based web browser due to be released from the
Googleplex on September 2, has been preceded by a lengthy comic book
explaining the principles behind its design. "But, when you have to
do interpretation, you have to look at the structure of your internal
representation over and over again. So instead, V8 looks at the JavaScript
source code and generates machine code that can run directly on the CPU
that's running the browser."
(Log in to post comments)
Well done Posted Sep 2, 2008 13:47 UTC (Tue) by proski (subscriber, #104) [Link] Let's hope the quality of the code is just as good as that of the comics.
Well done Posted Sep 2, 2008 13:50 UTC (Tue) by vblum (guest, #1151) [Link]
the best browser advertisement (most informative yet funny) I have ever seen. I didn't think I'd try it, but now I will.
(although FF3 was already a rather nice improvement for me)
Well done Posted Sep 2, 2008 14:38 UTC (Tue) by Zenith (subscriber, #24899) [Link]
AWESOME presentation!! :D
I am usually indifferent or only slightly interested in such presentations usually, but this rocks my socks off! Will definitely be trying it out.
One thing though: Is this supposed to be a tongue-in-cheek reference to Steve Jobs: http://www.google.com/googlebooks/chrome/images/37.jpg?
See the guy with the globe sneaking off :P
What future for Firefox Posted Sep 2, 2008 14:48 UTC (Tue) by rvfh (subscriber, #31018) [Link]
AFAIK, Google are the sponsors behind Firefox... But now they've re-developed a browser from scratch, and what's more it's based on Webkit and not Gecko (as FF is).
It would be nice to know whether Google intends to support two competitive projects, or... else? Anyone? Chris?
What future for Firefox Posted Sep 2, 2008 15:27 UTC (Tue) by salimma (subscriber, #34460) [Link]
Nokia's doing that to a certain extent too: their MicroB browser for the Internet Tablet is Gecko-based, but they were the main sponsor behind WebKit's GTK binding, and now owns Trolltech which has WebKit bindings.
What future for Firefox Posted Sep 2, 2008 15:37 UTC (Tue) by njs (guest, #40338) [Link]
Google just renewed their contract with Mozilla Corp. last week, and for a 3 year period this time instead of the 2 year renewal period they've been using to date. So apparently Google *will* be supporting two competitive projects until 2011, at least; plus the timing suggests that this is an intentional decision on Google's part.
What future for Firefox Posted Sep 2, 2008 23:59 UTC (Tue) by qg6te2 (guest, #52587) [Link] If the goal is to reduce the influence of Internet Exploder (and hence by extension the influence of MS), it makes sense for Google to support two browser projects. Instead of people having only two viable choices (i.e. IE and Firefox), three is better, simply as IE is now outnumbered with viable competition. Some people may not like the new interpretation of tabbing in Chrome (or they might be uncomfortable with "new" features) and hence they would stick to either IE or Firefox. On the other hand, people who do not like the "standard" tabbing approach in IE or Firefox (or are comfortable with "new" features) may go for the Chrome browser. This is somewhat akin to a company releasing the same product (say, potato chips) under two different brands, priced differently. Some people will think that the higher priced chips are "better" and hence would buy these, while other people are on a budget and hence would go for the lower priced alternative. Either way the company profits at the expense of the competition.
What future for Firefox Posted Sep 3, 2008 2:02 UTC (Wed) by ofeeley (guest, #36105) [Link]
And also will the CAcert root certificate be distributed with Google Chrome? Given Google's willingness to provide a list of bogus sites I wonder will they also help to end the commercial SSL certificate racket?
Well done Posted Sep 2, 2008 21:10 UTC (Tue) by nix (subscriber, #2304) [Link]
Seconded. This work of art makes even the GNOME release notes look bad by
comparison. (When was the last time you could call release notes a work of art? When was the last time you found someone describing crummy plugin security amusing?)
Well done Posted Sep 3, 2008 12:35 UTC (Wed) by k8to (subscriber, #15413) [Link]
Amusingly, the comic avoids traditional web links, using javascript in such a way that the back button does not work correctly for me, and the Next link silently does nothing at the end of the comic.
If this sort of overengineered yet buggy mindset is going into chrome...
Well done Posted Sep 3, 2008 13:16 UTC (Wed) by markhb (guest, #1003) [Link] well, considering that Google hired the guy who literally wrote the book on graphic literature, it's not surprising that the comic is excellent. I'll see how the browser is momentarily...
The Google Chrome comic book Posted Sep 2, 2008 14:00 UTC (Tue) by rfunk (subscriber, #4054) [Link]
Apparently they're starting with a Windows beta, and Linux/Mac versions
will come later. But it will be open-source. (I wonder when we'll get the source.) http://googleblog.blogspot.com/2008/09/fresh-take-on-brow...
The Google Chrome comic book Posted Sep 2, 2008 16:48 UTC (Tue) by lmb (subscriber, #39048) [Link]
No Linux? Doh. Wonderful. But with source, one should hope some kind soul provides a compilation ...
The Google Chrome Source Posted Sep 2, 2008 20:46 UTC (Tue) by rfunk (subscriber, #4054) [Link]
Turns out the development site is at: http://code.google.com/chromium/
and the code is at: http://src.chromium.org/svn/trunk/src/chrome/
Still Windows-only for now, though there are build instructions for Linux.
The Google Chrome Source Posted Sep 3, 2008 5:30 UTC (Wed) by SimonKagstrom (subscriber, #49801) [Link]
Apparently building for Linux only gives you a command line test executable:
http://dev.chromium.org/developers/how-tos/build-instruct...
So some work still until we'll really be able to do browsing with this :-)
// Simon
Javascript speed Posted Sep 2, 2008 14:03 UTC (Tue) by epa (subscriber, #39769) [Link]
It seems like every week there is an announcement of a new Javascript engine which will make everything so much faster than before and will be included in $Browser real soon now. Are there any meaningful benchmarks of Javascript performance compared to say Perl, Python, C# and Java? Should people start using Javascript for general scripting tasks and applications outside the browser, if it's really so fast?
Javascript speed Posted Sep 2, 2008 14:34 UTC (Tue) by tjc (subscriber, #137) [Link]
> Should people start using Javascript for general scripting tasks and applications outside the browser, if it's really so fast?
I'd like that, although I don't think it will ultimately be any faster than any other language with similar features. But a lot of people already know it, since it's the only practical choice for client-side web programming, and it would be nice to be able to write server-side scripts in the same language.
Javascript speed Posted Sep 2, 2008 15:25 UTC (Tue) by rfunk (subscriber, #4054) [Link] There are standalone javascript interpreters that you can run on your server. KDE has kjscmd, and Mozilla has Spidermonkey and (if you want to run JS on the JVM) Rhino.
They probably should Posted Sep 2, 2008 17:12 UTC (Tue) by khim (subscriber, #9252) [Link] Actually the fact that there are a lot of JITs and different implementations means it'll be fastest scripting language in a few years. People are doing work in parallel - and it's more-or-less at the point which took more the decade for python/perl/whatever and still is not ready to be really used in production. Now they'll start to pull ideas from implementations supplied by others... Hopefully someone will rip these V8 engine and put it in some small binary usable in shebang script's line...
Javascript speed Posted Sep 3, 2008 0:46 UTC (Wed) by gouyou (subscriber, #30290) [Link]
I would much prefer everybody integrating a nicer language like python in the browser. Javascript is useful but the syntax and libraries are dreadful.
Javascript speed Posted Sep 3, 2008 4:27 UTC (Wed) by flewellyn (subscriber, #5047) [Link]
On libraries, I'll agree...somewhat. It's only recently been taken seriously as a language, so library support is uneven. But I rather like Javascript's semantics; it's like a curly-braced Lisp.
Javascript speed Posted Sep 3, 2008 20:09 UTC (Wed) by robertm (subscriber, #20200) [Link]
Only without all the parts that make Lisp different from all the other scripting languages out there: a useful code/data correspondence; lexically scoped non-local returns; an object system based on generic functions instead of message passing; resumable exceptions; dynamically-scoped variables.
Javascript speed Posted Sep 3, 2008 21:40 UTC (Wed) by flewellyn (subscriber, #5047) [Link]
Fair enough. Still, it's got closures and good function/object polymorphism.
Javascript speed Posted Sep 4, 2008 8:05 UTC (Thu) by nix (subscriber, #2304) [Link]
And macros. Don't forget macros. (I suppose this is a consequence of
the 'useful code/data correspondence' that you did mention.)
Javascript speed Posted Sep 3, 2008 19:00 UTC (Wed) by tjc (subscriber, #137) [Link]
It could be a *lot* worse. As a Python programmer you should at least appreciate the fact that semicolon statement terminators are optional.
Myself, I like the first-class funtions, and the somewhat oddball prototype-based objects. I'm less thrilled with the way it silently converts objects from one type to another, in sometimes unexpected ways, but most scripting languages do this to some extent.
The Google Chrome comic book Posted Sep 2, 2008 14:11 UTC (Tue) by Sutoka (guest, #43890) [Link]
How much of the nifty stuff in Chrome is modification to Webkit (which'll hopefully make it back up stream), and how much are changes made above Webkit? Hopefully a lot of the changes (if they're actually useful) can migrate upstream into Webkit proper so all the other Webkit based browsers can get any of the nice stuff mostly for free...
The Google Chrome comic book Posted Sep 2, 2008 14:15 UTC (Tue) by rfunk (subscriber, #4054) [Link]
WebKit is just an HTML rendering engine. Most of the Chrome features
appear to be outside of that, with the possible exception of the part that makes plugins more independent.
The Google Chrome comic book Posted Sep 2, 2008 16:08 UTC (Tue) by lambda (subscriber, #40735) [Link]
Well, Chrome is many things. One is that it's a browser, with a new security and performance
model based on separate processes for each page renderer and JavaScript interpreter. This is beyond the scope of WebKit itself, so it won't make it up; Apple may decide to adopt some ideas in Safari, but that's closed source and beyond the scope of the WebKit project.
Another thing new in Chrome is Google's Windows WebKit port. Apple already has a Windows
Another new piece is Google's JavaScript JIT compiler, V8. They claim that it's a lot faster than
There may be other bits and pieces that could be used upstream, like the integrated Google
The Google Chrome comic book Posted Sep 2, 2008 15:56 UTC (Tue) by njs (guest, #40338) [Link]
One thing I'll be curious to see is how Google goes about developing this as FOSS -- and in particular, how and whether they end up building a development community outside Google itself. This is by far the largest FOSS codebase I can think of to come out of Google -- otherwise there's what, a few utility libraries for C++ and Python, Gears, ...?
If you look at the Gears mailing list, for example, it's a bunch of messages of the form "[user] has used [internal google tool] to request a code review! please go to [location on google intranet] to see the patch", and there are no commits from outside Google. Which is understandable, given all the comfy development tech that Google engineers have access to, and not really a problem for something like Gears... but one suspects that Chrome is going to have much more interest from the outside. (E.g., Gears still isn't available for Linux x86-64; apparently no-one cares enough to fix it. I doubt the FOSS crowd will let that kind of thing stand for Chrome.)
Interesting times!
If you mean "largest monolithic piece of code" then yes Posted Sep 2, 2008 17:16 UTC (Tue) by khim (subscriber, #9252) [Link] There are over 100 different libraries open-sourced by Google (and don't forget about SoC) so and I think they all combined are bigger then Chrome codebase, but yes, it's probably the largest monolithic piece of code ever open-sourced by Google (though Android probably rival that - but it's not open-source... yet).
If you mean "largest monolithic piece of code" then yes Posted Sep 3, 2008 5:11 UTC (Wed) by njs (guest, #40338) [Link]
Basically I mean "largest potential development community" -- the vast majority of Google code releases are API demos, little libraries written by 1 or 2 people, etc. You know, the sort of code that doesn't attract much external development whether it came from Google or just got posted on SourceForge. (And SoC is irrelevant, since the whole motivation for how it's put together is that Google didn't want to get into the business of building development communities, they wanted to leverage the existing experts, i.e. FOSS projects.)
Chrome OTOH has both a large and active team inside Google, and is likely to attract a large number of interested developers, so it seems to create a new challenge for them -- can they build a viable community of external contributors, and collaborate effectively with them?
And yeah, Android will be interesting to watch in this regard as well.
(...WTH did Google reimplement nscd? http://code.google.com/p/gnscd)
If you mean "largest monolithic piece of code" then yes Posted Sep 4, 2008 19:28 UTC (Thu) by pphaneuf (guest, #23480) [Link] Have you looked at nscd? ;-)
If you mean "largest monolithic piece of code" then yes Posted Sep 4, 2008 23:23 UTC (Thu) by nix (subscriber, #2304) [Link]
I've looked at glibc nscd. Neat, fast, efficient, config file format and
means of operation entirely undocumented (who needs network serialization when you can just shared-mmap() sparse files and read directly from the deserialized representation of the service files).
It's enormously more efficient than the Solaris ncsd used to be back in
The Google Chrome comic book Posted Sep 3, 2008 8:12 UTC (Wed) by sean.hunter (guest, #7920) [Link]
Perhaps this is a sign that I am becoming an old giffer, but I hated this comic book presentation. It may well be a great browser and I will certainly try it, but for all the technical innovations claimed, for me this announcement is the apotheosis of everything wrong with google.
For starters, this document is a set of static images. Why, then, does it not load if you have javascript disabled?
The Google Chrome comic book Posted Sep 3, 2008 15:07 UTC (Wed) by jordanb (guest, #45668) [Link]
You're not missing anything. The whole comic is a smallish essay stretched across 40 pages by drawing a talking head to say each sentence.
The Google Chrome comic book Posted Sep 3, 2008 16:37 UTC (Wed) by johnkarp (subscriber, #39285) [Link]
My take is that they wanted to give a presentation online, with lots of visual aids, etc. Not a text dump. The 'traditional' way of doing so is to make a powerpoint presentation and dump the slides online, or videorecord some talking head going through a powerpoint. Which are both extremely painful to watch, and don't work well on the web.
I'd say the format they used is about perfect for 'canning' a presentation for online distribution. Sure, they could have edited a presentation into a movie, a la 'An Inconvenient Truth', but you wouldn't be able to print that out, or watch it in most offices without a pair of headphones.
No matter the effectiveness of the medium is though, I doubt there will be much more like this. Most organizations aren't willing to use something so labor-intensive to produce.
The Google Chrome comic book Posted Sep 3, 2008 19:38 UTC (Wed) by ofeeley (guest, #36105) [Link] But what's wrong with hooking up the "prev" and "next" link at the bottom of each page to the appropriate page? That way the back button still works and again there's no need for javascript. As it is the presentation is a classic piece of Obtrusive Javascript. What's the point of writing their own goPage() function in javascript when a browser can already handle previous and next? I like the comic, I think it's a great presentation, but I'm baffled as to why they've done the above.
The Google Chrome comic book Posted Sep 3, 2008 23:15 UTC (Wed) by nix (subscriber, #2304) [Link]
Quite so. Plus, it's not some generic talking head: it's the Chrome
developers. :)
Window manager ideas Posted Sep 2, 2008 16:21 UTC (Tue) by dmarti (subscriber, #11625) [Link] One process per page looks like a sensible idea.The use of multiple processes in Chrome is just begging to have SE Linux support added. -- Russell Coker From a UI POV, though, itlooks like a tabbed window manager that only supports browser windows. Would be interesting to let different programs each have a tab in a top-level window.
Window manager ideas Posted Sep 2, 2008 20:30 UTC (Tue) by jwb (guest, #15467) [Link]
Yes, and it decorates its own window border, which means it will never fit in with the look and feel of, say, Ubuntu, Kubuntu, and Xubuntu all at once. This is a bad idea and it would be nice if software vendors would stop doing it. Private window decorations are one of the most glaring problems with Java GUI applications.
Window manager ideas Posted Sep 3, 2008 4:48 UTC (Wed) by njs (guest, #40338) [Link]
What do you mean? I haven't seen the Linux version, but the screenshots on the website for the windows version clearly show the standard Windows decorations. The only difference is that the tabs go above the address bar instead of below. Or are you saying that it's doing some games under the hood for some reason?
Window manager ideas Posted Sep 3, 2008 13:44 UTC (Wed) by djthomp (guest, #14080) [Link]
I would guess that jwb is talking about how the tabs are up in the title bar.
Also, it has what I assume is the Vista look (Aero, I think?), but it has it even on XP. So it doesn't appear to be using the platform window decorations at all, it must be drawing everything all on its own.
Window manager ideas Posted Sep 3, 2008 13:50 UTC (Wed) by dw (subscriber, #12017) [Link]
I believe jwb is referring at least to Chrome on Windows XP, where it looks like this:
http://dmw.me.uk/this-is-not-windows.png
Vista + "glass" looks like this:
http://dmw.me.uk/this-is-windows-vista.jpg
This is the only annoying thing I've found about Chrome thus far. Applications that customize the Windows UI, especially when it only yields something like an extra 6 vertical pixels, annoy the hell out of me.
Window manager ideas Posted Sep 3, 2008 15:09 UTC (Wed) by proski (subscriber, #104) [Link] As far as I know, Windows users are used to such things.
Window manager ideas Posted Sep 3, 2008 15:58 UTC (Wed) by jwb (guest, #15467) [Link]
No, it is not the standard windows decorations. The window borders are different colors and different sizes, and the title bar is kinda like the Vista title bar, but not really. It only looks approximately correct on Vista with Aero enabled. On XP it looks totally wrong, and on Windows 2000 it looks absurd.
Window manager ideas Posted Sep 3, 2008 23:06 UTC (Wed) by njs (guest, #40338) [Link]
Ah. Shows what I know about Windows :-)
Window manager ideas Posted Sep 3, 2008 0:55 UTC (Wed) by bojan (subscriber, #14302) [Link]
> One process per page looks like a sensible idea.
Use the fork(), Luke. ;-)
The Google Chrome comic book - how about the new html tags? Posted Sep 2, 2008 18:59 UTC (Tue) by riba (guest, #50370) [Link]
<video>, <audio> are they going to be supported by Chrome? The main advance in browsing we're hoping so far is the end of the flash requirement to watch videos.
The Google Chrome comic book Posted Sep 2, 2008 21:23 UTC (Tue) by tjc (subscriber, #137) [Link]
Google says:
> We've used components from Apple's WebKit and Mozilla's Firefox, among others [snip]
http://www.google.com/chrome/intl/en/why.html?hl=en
Webkit's contribution is obvious, but I wonder what they took from Firefox?
From Mozilla: Chrome
From Mozilla: Chrome Posted Sep 3, 2008 1:52 UTC (Wed) by tjc (subscriber, #137) [Link]
I thought the name sounded familiar. I even bought the book (Rapid Application Development with Mozilla). It (Mozilla Chrome) was really unstable a few years ago, so I gave up on it.
What Chrome took from Firefox Posted Sep 3, 2008 8:29 UTC (Wed) by darwish07 (subscriber, #49520) [Link] Maybe the experience of developing a widely used web browser, building above what's done good and improving what's done bad?
The Google Chrome comic book Posted Sep 3, 2008 9:24 UTC (Wed) by sylware (guest, #35259) [Link]
Chrome, if based on WebKit, is then (L)GPL, not BSD because WebKit is (L)GPL.
So, either Chrome is not based on webkit, either we face a GPL violation.
The Google Chrome comic book Posted Sep 3, 2008 9:39 UTC (Wed) by jospoortvliet (subscriber, #33164) [Link]
No, it can use the WebKit library, as it's LGPL, and the app itself can be
any other license. That's what the LGPL allows.
The Google Chrome comic book Posted Sep 3, 2008 10:05 UTC (Wed) by kripkenstein (subscriber, #43281) [Link]
Yes, WebKit is LGPL, Chrome itself is BSD, and it also uses a list of other FOSS libraries, some from Mozilla (which are therefore MPL/GPL/LGPL), and a few other libraries with various permissive licenses as well (MIT, zlib, etc.). See here:
http://code.google.com/chromium/terms.html
The Google Chrome comic book Posted Sep 3, 2008 13:55 UTC (Wed) by sylware (guest, #35259) [Link]
Then, chrome is the GUI connected to webkit.
Fine, they will have to publish any improvements made on webkit that will benefit all other GUIs based on this web engine. (I wish webkit would be coded in C)
The Google Chrome comic book Posted Sep 3, 2008 23:13 UTC (Wed) by nix (subscriber, #2304) [Link]
No. WebKit is an HTML rendering engine. The JS interpreter, plugin layer,
multiprocess tabbing, and the entire user interface that you interact with is *not* part of WebKit.
(and we know about your C fetish, thanks. Actually, it's not a C fetish:
The Google Chrome comic book Posted Sep 4, 2008 0:45 UTC (Thu) by robert_s (subscriber, #42402) [Link]
"(I wish webkit would be coded in C)"
I remember when I was a C++-o-phobe.
The Google Chrome comic book Posted Oct 5, 2008 15:51 UTC (Sun) by sylware (guest, #35259) [Link]
I remember when I was a C++-o-phil... a few projects latter and some C++ code reviews on some big projects... back to C!
C++ is too incentive for brain damaged design and the complexity of a C++ compiler is too insane to be reasonable (we have already the kernel and an optimizing C compiler and that's fairly enough!). It's a matter of reasonably keeping the software stack as low as possible in size and complexity. Each person has its own perception of the limits regarding the latter issue, and C++ is overboard as far as I am concerned.
The Google Chrome comic book Posted Sep 4, 2008 14:06 UTC (Thu) by pphaneuf (guest, #23480) [Link] Uh, the whole thing is open source, so they are not only publishing back their improvements to Webkit, but everything else as well! If the Firefix community wants to put V8 into their browser, they are welcome to do so. As for C/C++, what would be your guess on which language Brian Kernighan or Ken Thompson use for system programming nowadays?
The Google Chrome comic book Posted Sep 3, 2008 9:42 UTC (Wed) by engla (guest, #47454) [Link]
Google has a vision for a Greater German Reich(?), or at least the artist has put in some interesting own geography into the European map;
http://www.google.com/googlebooks/chrome/images/13.jpg
Germany has annexed all of Switzerland, Austria, Hungary, Slovenia, the whole Be-Ne-Lux and Denmark?
Apart from the new Germany, most other borders seem fine.
The Google Chrome comic book Posted Sep 3, 2008 13:00 UTC (Wed) by epa (subscriber, #39769) [Link]
I think the joke is that Denmark has annexed Germany, etc.
The Danish Reich Posted Sep 4, 2008 6:14 UTC (Thu) by ncm (subscriber, #165) [Link]
Good one. I would have missed that.
How to download? Posted Sep 3, 2008 10:30 UTC (Wed) by i3839 (guest, #31386) [Link]
Anyone any idea how to download the Windows version while running Linux?
It's nice that the download page links to a page which makes it possible to subscribe for an email when the Linux version is ready, but making it totally impossible to download the Windows version from Linux seems a bit weird.
Edit: Never mind, direct url seems to be http://dl.google.com/update2/installers/ChromeSetup.exe
AdBlock and other extensions Posted Sep 3, 2008 17:49 UTC (Wed) by pflugstad (subscriber, #224) [Link]
Initial impression is pretty good, but I really really *really* miss Ad Block Plus. Every time I'm forced to use a browser without it, I cringe at what a ghetto of crap advertising most of the Internet is.
Since it's open source, but it's also Google, it'll be interesting if Google accepts an ad blocking extension to it.
Along those, an extension mechanism like Firefox would likely push the adoption a lot. Even more if they can make those extensions (at least partly) compatible with Firefox.
BSD Posted Sep 7, 2008 14:13 UTC (Sun) by kbob (guest, #1770) [Link]
Let me be the first to predict that Chrome's BSD license will come back to bite Google in the butt when Microsoft forks off Microsoft Chrome, "Now with 35% more incompatibility!!!"
It's not terribly likely, as the NIH is strong at Microsoft, but I did want to be first. (-:
|
|||||||||||||
