If you have a chance can you re-run the tests and make sure I didn't screw anything up? :)
-ross On Wed, Sep 10, 2014 at 11:04 AM, Shazron <[email protected]> wrote: > Ross - shoot. Arrgh I may have mixed up my test apps :( I should emblazon > "WKWebView" or "UIWebView" at the start of the app. Thank goodness for peer > review... > > On Wed, Sep 10, 2014 at 1:11 AM, Ross Gerbasi <[email protected]> wrote: > > > Here are the results of some tests on iOS8, hopefully they will be useful > > to someone! :) > > I need to do some more useful tests on DOM related stuff as it was > > mentioned here already JS speed is not usually the bottleneck in Cordova > > apps. Unless you're doing crazy crypto like some people :) > > > > *@Shazron:* I am not seeing IndexedDB support in UIWebView but maybe I am > > not enabling something I need to on the native side? For me HTML5 Test is > > only showing IndexedDB for Safari and WKWebView. As you can see I am > > getting a dramatic lower HTML5 Test score in UIWEbView vs WKWebView. > > > > > > *Tests run on an iPod Touch 5th Generation running iOS8 GA* > > > > iPod Touch iOS 8 *Safari* > > UserAgent - Mozilla/5.0 (iPod touch; CPU iPhone OS 8_0 like Mac OS X) > > AppleWebKit/600.1.4 (KHTML, like Gecko) Version/8.0 Mobile/12A365 > > Safari/600.1.4 > > Sunspider - 1825ms Overall > > HMTL5 Test - 440/555 > > - Seamless iFrame - No > > - 2D Graphics Path Support - Yes > > - Web GL 3D Graphics - Yes > > - IndexedDB - Yes > > - Objectstore ArrayBuffer Support - Yes > > > > iPod Touch iOS 8 *HomeScreen* > > UserAgent - Mozilla/5.0 (iPod touch; CPU iPhone OS 8_0 like Mac OS X) > > AppleWebKit/600.1.4 (KHTML, like Gecko) Mobile/12A365 > > SunSpider - 1801.0ms Overall > > HTML5 Test - 427/555 > > - Seamless iFrame - No > > - 2D Graphics Path Support - Yes > > - Web GL 3D Graphics - Yes > > - IndexedDB - No > > - Objectstore ArrayBuffer Support - No > > > > > > iPod Touch iOS 8 *WKWebView* > > UserAgent - Mozilla/5.0 (iPod touch; CPU iPhone OS 8_0 like Mac OS X) > > AppleWebKit/600.1.4 (KHTML, like Gecko) Mobile/12A365 > > SunSpider - 1822.8ms Overall > > HTML5 Test - 438/555 > > - Seamless iFrame - No > > - 2D Graphics Path Support - Yes > > - Web GL 3D Graphics - Yes > > - IndexedDB - Yes > > - Objectstore ArrayBuffer Support - No > > > > > > iPod Touch iOS 8 *UIWebView* > > UserAgent - Mozilla/5.0 (iPod touch; CPU iPhone OS 8_0 like Mac OS X) > > AppleWebKit/600.1.4 (KHTML, like Gecko) Mobile/12A365 > > SunSpider - 7114.6ms Overall > > HTML5 Test - 427/555 > > - Seamless iFrame - No > > - 2D Graphics Path Support - Yes > > - Web GL 3D Graphics - Yes > > - IndexedDB - No > > - Objectstore ArrayBuffer Support - No > > > > > > *iOS 8 Notes* > > - Seamless iFrame was removed compared to iOS 7.1.2 > > - IndexedDB is only supported in Safari & WKWebView App > > - Objectstore ArrayBuffer is only supported in Safari & WKWebView App > > - UIWebView is ~4x slower then WKWebView for JS (Sunspider results) > > - WKWebView cannot load local files (issue for Cordova/Phonegap/Offline > > Solutions) > > - No UserAgent difference between UIWebView and WKWebView. > > > > > > On Tue, Sep 9, 2014 at 7:41 PM, tommy-carlos williams < > [email protected]> > > wrote: > > > > > Generally speaking, JS performance is not a problem for most iOS > Cordova > > > apps. Certainly I never had a problem until recently. > > > > > > Sadly, I now work with a JS privacy (crypto) lib called Crypton[1] that > > > works pretty damn good on Android, decently on BB10 and even pretty > good > > on > > > FxOS… but takes like 2 MINUTES just to log in on an iPhone 5 :( > > > > > > And that’s only using 1000 rounds of PBKDF2 for SRP authentication > (less > > > than we would like). > > > > > > Luckily, my use case is not the norm… but it does mean I am making apps > > > that are unusable on iOS, which makes me sad. > > > > > > 1. https://crypton.io > > > > > > On 10 September 2014 at 10:20:21, Andrew Grieve ([email protected]) > > > wrote: > > > > > > That's great news. > > > > > > JS execution speed honestly hasn't really been a problem for Cordova > > > apps that I've heard of anyways. > > > > > > On Tue, Sep 9, 2014 at 7:01 PM, Shazron <[email protected]> wrote: > > > > Some interesting findings - iOS UIWebView and WKWebView have the same > > > > user-agent value, and both score 440 on html5test.com. > > > > > > > > Which makes me think they use the same engine (renderer and js), but > > > > UIWebView has nitro turned off. WKWebView seems to be about 4x faster > > on > > > > the Sunspider benchmark. > > > > > > > > On Tue, Sep 9, 2014 at 3:58 PM, Shazron <[email protected]> wrote: > > > > > > > >> UIWebView iOS 8 scores 440/555 points on html5test.com, and yes it > > > >> supports indexedDB > > > >> Full results: http://html5te.st/f16d892273bbe5bd > > > >> > > > >> On Tue, Sep 9, 2014 at 1:33 PM, Andrew Grieve <[email protected] > > > > > >> wrote: > > > >> > > > >>> :( > > > >>> > > > >>> Hopefully UIWebView will still get some updates though? E.g. > > IndexedDb? > > > >>> > > > >>> On Tue, Sep 9, 2014 at 4:18 PM, Tommy Williams <[email protected] > > > > > >>> wrote: > > > >>> > That's pretty disappointing. > > > >>> > > > > >>> > Especially with not even a comment from apple. > > > >>> > On 10 Sep 2014 05:50, "Shazron" <[email protected]> wrote: > > > >>> > > > > >>> >> Well, bad news. I just installed the iOS 8 GM firmware on an > > iPhone > > > 5 > > > >>> and > > > >>> >> used Xcode 6 GM, and the file:// url "bug" (assuming it is a bug > > and > > > >>> not by > > > >>> >> design) is still there. > > > >>> >> > > > >>> >> http://www.openradar.me/radar?id=5839348817723392 > > > >>> >> > > > >>> >> On Fri, Sep 5, 2014 at 12:48 PM, Shazron <[email protected]> > > wrote: > > > >>> >> > > > >>> >> > Clarification: > > > >>> >> > When I said "The bridge works great, and plugins work great." > > this > > > >>> was > > > >>> >> > for loading a html page and plugins using the file:// protocol > > > using > > > >>> >> > WKWebView (as the title of the report implies). The bug I > > reported > > > >>> was on > > > >>> >> > Device (> iOS beta 4), on Simulator it is *fine* (this info > was > > in > > > >>> the > > > >>> >> bug > > > >>> >> > report as well). > > > >>> >> > > > > >>> >> > *Nothing* has been done using the local web server and proxy > (at > > > >>> least > > > >>> >> > nothing checked in). An implementation on how the proxy works > > can > > > be > > > >>> seen > > > >>> >> > in the PhoneGap Developer App: > > > >>> >> > https://github.com/phonegap/phonegap-app-developer > > > >>> >> > > > > >>> >> > See CB-7043 for progress on tasks regarding this. > > > >>> >> > > > > >>> >> > > > > >>> >> > > > > >>> >> > On Fri, Sep 5, 2014 at 11:40 AM, Shazron <[email protected]> > > > wrote: > > > >>> >> > > > > >>> >> >> I figure I will write this all up before the official release > > of > > > >>> iOS 8 > > > >>> >> >> next week (probability high) and everyone asking about > support. > > > >>> >> >> > > > >>> >> >> It has stalled because the WKWebView cannot load files using > > the > > > >>> file:// > > > >>> >> >> protocol since iOS 8 beta 4. > > > >>> >> >> > > > >>> >> >> This bug has been filed with Apple weeks ago: > > > >>> >> >> http://www.openradar.me/radar?id=5839348817723392 > > > >>> >> >> > > > >>> >> >> I even checked WebKit check-ins if there was any progress, so > > > far, > > > >>> no: > > > >>> >> >> > > > >>> >> >> > > > >>> >> > > > >>> > > > > > > http://trac.webkit.org/browser/trunk/Source/WebKit2/UIProcess/API/Cocoa?order=date&desc=1 > > > >>> >> >> (but it's entirely possible the loading code is in another > part > > > of > > > >>> the > > > >>> >> >> tree). > > > >>> >> >> > > > >>> >> >> The alternative is to run a local web server, which works > > great. > > > >>> >> However, > > > >>> >> >> this will open up a can of worms possibly with Apple, I'm not > > > sure. > > > >>> >> >> > > > >>> >> >> The other interesting tidbit is, with WKWebView, for locally > > > loaded > > > >>> >> files > > > >>> >> >> using the file:// protocol, cross-domain restrictions now > > apply, > > > >>> unlike > > > >>> >> >> UIWebView's behaviour. To have the same behaviour as > UIWebView, > > > we > > > >>> would > > > >>> >> >> need to proxy these requests (modify xhr.open to go to our > > proxy, > > > >>> which > > > >>> >> >> requires the local web server). > > > >>> >> >> > > > >>> >> >> The bridge works great, and plugins work great. > > > >>> >> >> > > > >>> >> >> > > > >>> >> >> > > > >>> >> >> > > > >>> >> >> > > > >>> >> > > > > >>> >> > > > >>> > > > >> > > > >> > > > > > > > > >
