FullscreenExitBubble temp UI for Mac.

chrome/browser/ui/cocoa/browser_window_controller.mm

Index: chrome/browser/ui/cocoa/browser_window_controller.mm
diff --git a/chrome/browser/ui/cocoa/browser_window_controller.mm b/chrome/browser/ui/cocoa/browser_window_controller.mm
index 6444bbce94fb7c45801a7a0a15062afe12f7b5b8..5aa2e0bb6190dfa1a60fecdbf9396206921b36cf 100644
--- a/chrome/browser/ui/cocoa/browser_window_controller.mm
+++ b/chrome/browser/ui/cocoa/browser_window_controller.mm
@@ -1948,13 +1948,36 @@ willAnimateFromState:(bookmarks::VisualState)oldState
     enteredPresentationModeFromFullscreen_ = YES;
     if ([[self window] isKindOfClass:[FramedBrowserWindow class]])
       [static_cast<FramedBrowserWindow*>([self window]) toggleSystemFullScreen];
-    return;
+  } else {
+    if (fullscreen)
+      [self enterFullscreenForSnowLeopardOrEarlier];
+    else
+      [self exitFullscreenForSnowLeopardOrEarlier];
   }
 
-  if (fullscreen)
-    [self enterFullscreenForSnowLeopardOrEarlier];
-  else
-    [self exitFullscreenForSnowLeopardOrEarlier];
+  // Disable the floating tab bar if this fullscreen request was from a page.
+  // It will be re-enabled by the FullscreenExitBubble after it disappears.
+  BOOL fullscreen_for_tab = browser_->is_fullscreen_for_tab();

[Nico, 2011/09/15 04:37:32]

Ah. Maybe "allow_tab_switching_in_fullscreen()"?

[jeremya, 2011/09/15 06:33:28]

That's not quite it -- we still allow tab switching, but if in content-initiated
fullscreen, a tab switch will exit fullscreen (if the window wasn't in
fullscreen before the request).

When the exit bubble disappears, it will turn the floating chrome back on.

+  [self setShowFloatingChrome:fullscreen && !fullscreen_for_tab];
+
+  if (fullscreen && fullscreen_for_tab) {
+    fullscreenExitBubbleController_.reset(
+        [[FullscreenExitBubbleController alloc] initWithOwner:self
+                                                      browser:browser_.get()]);
+    NSView *contentView = [[self window] contentView];

[Nico, 2011/09/15 04:37:32]

space goes to the right of *, not left

[jeremya, 2011/09/15 06:33:28]

Done.

+    CGFloat maxY = [self placeBookmarkBarBelowInfoBar] ?
+      NSMinY([[toolbarController_ view] frame]) :
+      NSMinY([[bookmarkBarController_ view] frame]);

[Nico, 2011/09/15 04:37:32]

Move this ?: construct into a well-named method

[jeremya, 2011/09/15 06:33:28]

Done.

... actually, is there any reason I couldn't just use
NSMaxY([contentView frame])?

+    NSRect fsebFrame = [[fullscreenExitBubbleController_ animatableView] frame];

[Nico, 2011/09/15 04:37:32]

no abbreviations. bubbleFrame?

[Nico, 2011/09/15 04:37:32]

It feels like all this frame computation business could be moved into a method
on fullscreenExitBubbleController (hand in maxWidth and maxY)

[jeremya, 2011/09/15 06:33:28]

Done.

[jeremya, 2011/09/15 06:33:28]

Done.

+    CGFloat maxWidth = NSWidth([contentView frame]);
+    fsebFrame.origin.x = (int)(maxWidth/2-NSWidth(fsebFrame)/2);
+    fsebFrame.origin.y = maxY-NSHeight(fsebFrame);
+    [[fullscreenExitBubbleController_ view] setFrame:fsebFrame];
+    [contentView addSubview:[fullscreenExitBubbleController_ view]];

[Nico, 2011/09/15 04:37:32]

Have you considered making the bubble a child window instead? (not saying that
that's necessarily better, mostly curious why you went with a view. a child
window probably _is_ a bit more robust.)

[jeremya, 2011/09/15 06:33:28]

I hadn't seen that method until I saw the confirm_quit dialog -- this is what
findbar uses.

I'm not sure why using a window would be more robust, but I'm happy to refactor
the code that way if you think it'd be better :)

Also, I think this UI is probably simple enough to construct without a XIB, but
I'm not sure how to restructure the code to do that...

[Nico, 2011/09/15 20:09:14]

Most overlapping things are child windows (bookmark bubble, content settings
bubbles, page action bubbles, confirm-to-quit, tabpose, download shelf download
arrow that animates down, status bubble...). We had to go through huge
contortions to make the findbar paint on top of the web contents when there's
accelerated content on the web page; overlapping views don't work all that well.
If there's not a good reason to keep it as a subview, I'd prefer a child window.
It shouldn't require many changes.
The Sad Tab View doesn't use a xib I believe, you can look at that for how to do
that.

+  } else {
+    [[fullscreenExitBubbleController_ view] removeFromSuperview];
+    fullscreenExitBubbleController_.reset();

[Nico, 2011/09/15 04:37:32]

so non-js fullscreen won't get the bubble? wasn't that what bug 14471 was about?

[jeremya, 2011/09/15 06:33:28]

Correct. I don't think we need to notify the user if they pressed the button
themselves. It might be a good idea to get confirmation from glen on this.

[Nico, 2011/09/15 20:09:14]

Yes…email chrome-ui-leads, and cc rsesek. Just say something along the lines of
"We think we should show the bubble only for js fullscreen because
$some_lion_argument. ok?"

+  }
 }
 
 - (BOOL)isFullscreen {
@@ -2051,6 +2074,11 @@ willAnimateFromState:(bookmarks::VisualState)oldState
   [self layoutSubviews];
 }
 
+- (void)setShowFloatingChrome:(BOOL)showChrome {

[Nico, 2011/09/15 04:37:32]

Do you want to

  DCHECK(showChrome ^ showFloatingChrome_);

? If that's not true, things will be buggy (you'd need a counter in that case)

[jeremya, 2011/09/15 06:33:28]

Is layoutSubviews not idempotent?

If not, perhaps we could just return if showFloatingChrome_ == showChrome?

[Nico, 2011/09/15 20:09:14]

I was thinking of two callers that don't know of each other, both setting this
to yes and then just one setting it to no

+  showFloatingChrome_ = showChrome;
+  [self layoutSubviews];
+}
+
 - (BOOL)isBarVisibilityLockedForOwner:(id)owner {
   DCHECK(owner);
   DCHECK(barVisibilityLocks_);