We added a pass to GWT before to work around a Safari Nitro bug with the shift operator (>>), we could do something similar. See here: http://code.google.com/p/google-web-toolkit/source/browse/trunk/dev/core/src/com/google/gwt/dev/js/JsCoerceIntShift.java
On Wed, Feb 6, 2013 at 4:03 AM, Jens <jens.nehlme...@gmail.com> wrote: > Hi, > > I just wanted to start a discussion on all the WebKit issues that appeared > after the release of Safari 6 / iOS 6. To get an impression here are some > links I found so far: > > GWT Group: > https://groups.google.com/d/topic/google-web-toolkit/gh7NbjaSb-k/discussion > https://groups.google.com/d/topic/google-web-toolkit/Snj52ylNSF4/discussion > https://groups.google.com/d/topic/google-web-toolkit/hduYnZG8bws/discussion > https://groups.google.com/d/topic/google-web-toolkit/ZmMNVGWRkVU/discussion > > Vaadin: > http://dev.vaadin.com/ticket/10722 > http://dev.vaadin.com/ticket/10753 > > Stackoverflow: > > http://stackoverflow.com/questions/12534409/ios-6-javascipt-intermittent-issues-with-object-defineproperty > * > http://stackoverflow.com/questions/13147026/disabling-jit-in-safari-6-to-workaround-severe-javascript-jit-bugs > * > * > * > The GWT groups discussions are mainly dealing with random app crashes > caused by ClassCastExceptions. Often variables of type HashMap<String, ?> > are involved. Problems with HashMaps containing String keys are also > reported to Vaadin. > > The last bold link is the one I am stumbled upon yesterday and its pretty > interesting. It states that WebKit's JIT compiler has some serious bugs and > it also contains a working proof of concept demonstrating an issue where > var++ and var-- do not work correctly when JIT is active. > This happens on iOS 6 devices, WebKit nightly builds, Safari 6 in Mac OS > 10.7 and 10.8. So I would guess in most or even all WebKit based browsers > that have JIT enabled. > > As soon as you connect to an iOS device using Safari for remote debugging > or when you open the Safari DevTools on Mac OS the issues disappear. I > would assume that WebKit simply disables JIT as soon as a debugger runs and > thats why all issues disappear magically. This would also explain why > Google Chrome for iOS does not suffer from these issues because I think > only Safari has access to the JIT compiler on iOS for security reasons. > > Currently as soon as a customer uses Safari/WebKit (maybe except Chrome > because of V8) to access a GWT app it is not guaranteed to work. We just > tested it with our application on Mac OS 10.7 / 10.8 Safari 6 and > discovered that it indeed has strange random behavior if you only try it > often enough. In our case for example we have a Map and containsKey with a > String key randomly stops working which results in wrong information shown > in the UI. So before today we were under the impression that its an iOS bug > only (which is still present in iOS 6.1) but it seems like its more wide > spread. > > So whats your opinion on this? Should GWT try to workaround these issues > somehow? On Stackoverflow a suggested workaround is to disable JIT for a > given method by wrapping the methods content in a try catch block that > re-throws possible exceptions. Would it be possible to introduce a flag to > GWT that enables such specific "hacks" for Safari permutations? As it > sounds like that JIT compilation is the main issue its maybe worth it to > give it a try? > > The author on stackoverflow already has an Apple ticket (ID 12606761) and > we also have created a ticked for WebKit itself which needs to be confirmed > and some votes I guess: https://bugs.webkit.org/show_bug.cgi?id=109036 > > Do you guys at Google know these issues and working on it? Don't you have > problems with your own GWT apps? > > Currently we have to tell customers to not use Safari at all and use > Chrome app on iOS devices. The whole situation is somewhat frustrating. > > > -- J. > > -- > -- > http://groups.google.com/group/Google-Web-Toolkit-Contributors > --- > You received this message because you are subscribed to the Google Groups > "Google Web Toolkit Contributors" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to google-web-toolkit-contributors+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > > -- -- http://groups.google.com/group/Google-Web-Toolkit-Contributors --- You received this message because you are subscribed to the Google Groups "Google Web Toolkit Contributors" group. To unsubscribe from this group and stop receiving emails from it, send an email to google-web-toolkit-contributors+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
