Fix regression caused by and bug unmasked by bug 373122 patch (display

problem and keyboard focus problems). b=384343+383821 r=josh sr=roc
This commit is contained in:
smichaud%pobox.com 2007-06-21 23:58:16 +00:00
Родитель 322ea3207c
Коммит 8c4e0de81b
1 изменённых файлов: 14 добавлений и 11 удалений

Просмотреть файл

@ -3785,13 +3785,14 @@ static PRBool IsSpecialGeckoKey(UInt32 macKeyCode)
// This method is called when are are about to lose focus.
// We must always call through to our superclass, even when mGeckoChild is
// nil -- otherwise the keyboard focus can end up in the wrong NSView.
- (BOOL)resignFirstResponder
{
if (!mGeckoChild)
return NO;
if (mGeckoChild) {
nsFocusEvent event(PR_TRUE, NS_LOSTFOCUS, mGeckoChild);
mGeckoChild->DispatchWindowEvent(event);
}
return [super resignFirstResponder];
}
@ -3834,16 +3835,18 @@ static PRBool IsSpecialGeckoKey(UInt32 macKeyCode)
}
// If the call to removeFromSuperviewWithoutNeedingDisplay isn't delayed from
// nsChildView::TearDownView(), the NSView hierarchy might get changed during
// calls to [ChildView drawRect:], which leads to "beyond bounds" exceptions
// in NSCFArray. For more info see bmo bug 373122. Apple's docs claim that
// If the call to removeFromSuperview isn't delayed from nsChildView::
// TearDownView(), the NSView hierarchy might get changed during calls to
// [ChildView drawRect:], which leads to "beyond bounds" exceptions in
// NSCFArray. For more info see bmo bug 373122. Apple's docs claim that
// removeFromSuperviewWithoutNeedingDisplay "can be safely invoked during
// display" (whatever "display" means). But it's _not_ true that it can be
// safely invoked during calls to [NSView drawRect:].
// safely invoked during calls to [NSView drawRect:]. We use
// removeFromSuperview here because there's no longer any danger of being
// "invoked during display", and because doing do clears up bmo bug 384343.
- (void)delayedTearDown
{
[self removeFromSuperviewWithoutNeedingDisplay];
[self removeFromSuperview];
[self release];
}