REGRESSION: Yes SEVERITY: Is in the eye of the beholder We have a Cocoa Kiosk built using Webkit that creates the mainWindow using NSBorderlessWindowMask in level GSShieldingWindowLevel. We presently build with version 418.8 of webkit, XCode version 2.4, and find that Flash mouse events are lost when using NSBorderlessWindowMask. We have solved this problem temporarily using NSTitledWindowMask, but this is, of course, not a perfect solution. No Flash mouse events are lost with NSTitledWindowMask. We do have the mainWindow as an NSWindow subclass and have canBecomeKeyWindow returning YES. We make one simple NIB file and do most of this programmatically. Previous builds with Panther versions did not have this problem. As our product is a production level Assessment Testing Kiosk, we are generally reluctant to upgrade to latest versions without extensive compatiblity testing. So far, with our recent upgrades, the only big glitch we have hit is the one above. The Flash in question is created as Flash MX 2004, ActionScript 2.0, Player 7 documents. Our QA tests with Flash Player plugins from version 7 up. Since I can turn the problem on or off by toggling NSBorderlessWindowMask or NSTitledWindowMask, the bug points towards NSBorderlessWindowMask. We support and our QA tests with Panther and Tiger. Regards, Richard
WebKit 418.8 is not the latest version available for Tiger. Have you tried with 418.9.1? http://developer.apple.com/internet/safari/uamatrix.html Also, please note that WebKit Bugzilla is for fixing bugs on "tip-of-tree" (ToT) WebKit development. Bugs in previous versions of WebKit, if they are fixed on ToT, are usually closed as RESOLVED/WORKSFORME or RESOLVED/FIXED. Having said that, I suggest you do two things: 1. File a Radar bug on http://bugreport.apple.com/ regarding this issue, referencing this Bugzilla bug in the Radar bug. (Also, please report the Radar bug assigned back here and add the "InRadar" keyword to this bug.) If you don't have an ADC account, you may create a free "Online" account using http://connect.apple.com/. This bug report is what Apple will use if they decide to fix the bug in Tiger by creating another build of the WebKit 418.x series. 2. Please test this bug against ToT WebKit to determine if it's been fixed yet, and report back here. Build instructions may be found on http://webkit.org/. You may download nightly WebKit builds (http://nightly.webkit.org/) if you have an HTML test case that's easy to load into a browser. Otherwise, it may be easiest to simply build your own copy of the WebKit framework and link against it. (Attaching a test case to this bug--reduced or not--would also be helpful.)
Another thing to consider is Bug 12199 Comment #3. How did you handle this situation when an updated WebKit was provided for Panther that matched Tiger? Would you handle the situation the same way if it happened with Tiger/Leopard?
I have more information about this bug. I was researching a bug in my own fullscreen browser and came across this report. This is exactly what I'm seeing too - in both the latest WebKit that come with Leopard and also in the nightly webkit (I embedded the latest webkit frameworks in my app and tested out my pages). I used r29876. Details of what I've seen/tested: - on a page where I load some flash components in their own floating window I cannot move or interact with the components when they are displayed. Just as Richard observed it is like the events aren't getting to the flash component - or are getting there really slowly.. if I click a button ten times it might register once that it was clicked. - if I switch to a titled window there is no issue as Richard already observed. - I built and tested the app on 10.5.1 and then also built the WebKit and embedded it in my app and tested again. The issue is still there. - I also found something else sort of related too - on a dual monitor system with a full screen window on one monitor. I'd get to the flash view page. Id try to interact with the flash component and it would behave badly (as described above). Then I'd click on a different app/window on my secondary monitor and then back to the full screen app and the flash components would work just fine (with borderless window). I don't really have a test app I can upload and the URLs I'm using for testing are not public either. I'm going to be working on this though and try and figure out what is happening. If anyone from apple wants my test app and the url with a reproducible case please send email to tthomas at tsquaredsoftware dot com
<rdar://problem/5721453>
If this hasn't been fixed, we're unlikely to fix it. Closing.