REGRESSION (r191907): Maxthon Browser -After exit full screen video playback the browser window is blank with audio running
<rdar://problem/26674003>
Created attachment 283555 [details] Patch
Comment on attachment 283555 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=283555&action=review > Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:68 > +@interface NSView (WebNSViewDetails) > +- (NSArray *)_referencingConstraints; > +@end This should be in an SPI header. Actually... do you ever use it? > Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:254 > + self.savedConstraints = _webView.superview.constraints; Did you mean to use _referencingConstraints here? I'm not sure that saving all constraints on a view we don't own is right. > Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:317 > + [NSLayoutConstraint activateConstraints:self.savedConstraints]; Should we be more careful if this throws?
Created attachment 283556 [details] Patch
Comment on attachment 283555 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=283555&action=review >> Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:254 >> + self.savedConstraints = _webView.superview.constraints; > > Did you mean to use _referencingConstraints here? I'm not sure that saving all constraints on a view we don't own is right. Nope. I removed the SPI header in the most recent patch. >> Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:317 >> + [NSLayoutConstraint activateConstraints:self.savedConstraints]; > > Should we be more careful if this throws? Will this throw? that seems.... Weird.
Created attachment 283567 [details] Patch
Comment on attachment 283567 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=283567&action=review > Source/WebKit2/UIProcess/mac/WKFullScreenWindowController.mm:313 > + @try { I mean, if you're going to do it, do it to both :) And one would expect some {BEGIN,END}_BLOCK_OBJC_EXCEPTIONS, not this.
How are going to test this so it doesn't regress again?
Created attachment 283823 [details] Patch for landing.
Created attachment 283831 [details] Patch for landing.
(In reply to comment #8) > How are going to test this so it doesn't regress again? I'm working on an API test for this.
Comment on attachment 283831 [details] Patch for landing. Clearing flags on attachment: 283831 Committed r203335: <http://trac.webkit.org/changeset/203335>