News for hackers. Stuff that matters.
Re:Google Chrome now works on iOS by nthitz on Thursday June 28, 2012 @01:43PM (#4172858)
Does it use it's own rendering engine or is it just interface on top of a UIWebView?
Re:Google Chrome now works on iOS by mbrubeck on Thursday June 28, 2012 @02:15PM (#4173042)
Vivian Li Cromwell from the Chrome team confirms that it uses the system rendering engine (presumably via UIWebView) and does not include the V8 JIT:

https://twitter.com/viviancromwell/status/218402587760795648

Re:Google Chrome now works on iOS by mtgx on Thursday June 28, 2012 @02:50PM (#4173267)
Well that's unfortunate. And didn't we learn last year that Apple left WebView twice as slow as the browser? Or did they fix that?
Re:Google Chrome now works on iOS by SoftwareMaven on Thursday June 28, 2012 @03:08PM (#4173365)
UIWebView is slower for security reasons. (Non-Apple) Apps aren't allowed to mark memory as an executable code space, which JITs require. Perhaps there would be some way for Apple to solve that for Nitro without introducing security concerns for all apps, but I'm not convinced.

I much prefer knowing apps can't surreptitiously create new executable parts of themselves.

Re:Google Chrome now works on iOS by kylebrown on Thursday June 28, 2012 @03:46PM (#4173536)
I don't understand why they can't sandbox the UIWebView. That's how mobile Safari (presumably) is secured, as is Chrome. It has to render/execute untrusted content in any case.
Re:Google Chrome now works on iOS by TazeTSchnitzel on Thursday June 28, 2012 @04:56PM (#4173884)
Performance, probably. To sandbox it I imagine it would have to run in a separate process, and then you'd have to pass messages back and forth, which might be slow. But who knows.

It may be one of these things they were planning to do later, and then someone pointed out it would possibly be detrimental to what they want the iOS ecosystem to be, so they scrapped plans.

Re:Google Chrome now works on iOS by kylebrown on Thursday June 28, 2012 @05:57PM (#4174178)
Message passing between a UIWebView is already one of the big complaints, requiring slow and klunky workarounds. To receive a message you detect URL anchor (#foo) changes in objective C. Sending is a bit better (inject a js string).

I wonder if its possible to sandbox a thread as opposed to a separate process. I doubt that Apple consciously neglects HTML5 app SDK functions. iOS has been the best platform for HTML5 for a couple of years already, supporting features (CSS 3d for one) that Android didn't.

Re:Google Chrome now works on iOS by TazeTSchnitzel on Thursday June 28, 2012 @06:00PM (#4174193)
They probably don't mind neglecting it for apps.
Re:Google Chrome now works on iOS by Synaesthesia on Thursday June 28, 2012 @01:51PM (#4172907)
Apps on the store have to use UIWebView, which of course doesn't get the Nitro JavaScript engine.
Re:Google Chrome now works on iOS by gcp on Thursday June 28, 2012 @02:01PM (#4172965)
Does this mean it will be slow as molasses on JavaScript-heavy sites?
Re:Google Chrome now works on iOS by tubbo on Thursday June 28, 2012 @02:52PM (#4173280)
They "have to"? I was under the impression that you had a choice of UIWebView and your own custom implementation. Are they really going to block sales of an app if it has an alternative JS engine?
Re:Google Chrome now works on iOS by Synaesthesia on Thursday June 28, 2012 @03:12PM (#4173379)
You might be able to do your own JavaScript implementation, not sure, but I think actually not. (I stand corrected on this). But V8 definitely won't be permitted under the current laws, as it uses executable blocks of code in memory, which are not permitted, for security reasons under iOS.

Don't see that rule changing. Let's hope Apple let other browsers be the default though.

Re:Google Chrome now works on iOS by mbrubeck on Thursday June 28, 2012 @04:17PM (#4173697)
No, Apple's iOS developer agreements do not allow browsers with their own JavaScript engines:

"Interpreted code may only be used in an Application if all scripts, code and interpreters are packaged in the Application and not downloaded. The only exception to the foregoing is scripts and code downloaded and run by Apple’s built-in WebKit framework."

Re:Google Chrome now works on iOS by patrickaljord on Thursday June 28, 2012 @04:07PM (#4173644)
Isn't UIWebView a component written by Apple? Why not include Nitro in it? Why would it be a security issue on Chrome but not on Safari? Do people have access to the UIWebView source code? If not, this "security" reason is BS.
Re:Google Chrome now works on iOS by mbrubeck on Thursday June 28, 2012 @04:16PM (#4173693)
Third-party processes on iOS cannot mark data pages as executable, which is a great security feature but it prevents JIT compilers or any other technology that dynamically generates machine code. Apple's built-in apps have an exception to this security mechanism, presumably because Apple is more confident in their own security auditing of those apps. Perhaps with some sandboxing mechanism and/or code auditing process they could allow some third-party apps similar access in the future, but it's not clear whether they feel a need to.
Re:Google Chrome now works on iOS by patrickaljord on Thursday June 28, 2012 @07:32PM (#4174490)
> Third-party processes on iOS cannot mark data pages as executable

Doesn't the UIWebView do that for you? And isn't it written by Apple? Why couldn't Apple make it use Nitro? Not sure where the security risk would be.

Re:Google Chrome now works on iOS by mbrubeck on Thursday June 28, 2012 @11:36PM (#4175120)
The restriction on executable pages is enforced by the kernel at the process level. This particular mechanism won't allow for enforcing diffeerent policiees for UIWebView code running in the same process as app code. If UIWebView could run JavaScript code in a separate process like Chrome then it might be possible.
Re:Google Chrome now works on iOS by patrickaljord on Friday June 29, 2012 @12:13AM (#4175205)
I thought Safari implemented multi-process a long time ago, not on mobile?