Re: Scrollers on custom view appearing but not disappearing

FROM : Quincey Morris
DATE : Sat Mar 01 07:33:17 2008

On Feb 29, 2008, at 21:31, Steve Weller wrote:

> -(void)frameDidChangeNotification:(NSNotification *)notification
> {
>
>     [[NSNotificationCenter defaultCenter] removeObserver: self];
>     NSRect frame = [[self superview] frame];
>     NSSize cSize;
>     cSize = [self canvasSize];
>         
>     // Use the larger dimensions of the canvas and the superview
>     if(frame.size.width > cSize.width)
>         cSize.width = frame.size.width;    
>     if(frame.size.height > cSize.height)
>         cSize.height = frame.size.height;
>
>     [self 
> setFrame:NSMakeRect
> (frame.origin.x,frame.origin.y,cSize.width,cSize.height)];
>     //    NSLog(@"%f %f",newSize.width, newSize.height);
>     
>     NSNotificationCenter *center = [NSNotificationCenter defaultCenter] ;
>    [center addObserver: self
>               selector: @selector(frameDidChangeNotification:)
>                   name: NSViewFrameDidChangeNotification
>                  object: self];
>     
> }
>
> This figures out the rectangle that encloses both the canvas and the 
> superview's frame and makes my custom view's frame equal to that. 
> The only remaining thing to fix is that the lower left point is 
> always shown in the view, when I actually want the center point to 
> be shown. So I have to shift the frame origin as part of the 
> calculation.


A couple of small points:

-- It's not quite correct to use the superview's frame to calculate a 
view's frame, since they are in different coordinate systems. You 
really should use [[self superview] bounds], which is in the same 
coordinate system as [self frame].

The problem is harmless in this case, because the superview is a 
NSClipView, which happens to keep its frame coordinate system 
synchronized with that of view it contains, but this is not generally 
true of view-superview geometry.

-- If you want to refer to the clip view, [[self enclosingScrollView] 
contentView] is more correct than [self superview]. The fact that 
they're the same thing is an implementation detail. (But if you're 
going to pretend not to know they're the same you really should do an 
explicit coordinate conversion when combining their dimensions.)

-- As someone suggested on this list a few weeks ago, it's perhaps 
marginally more elegant to use [self visibleRect] instead of the clip 
view bounds. Although the purpose of the clip view is to manage the 
visible rect of the view it contains, using the visible rect directly 
means you don't have to build in knowledge of that implementation 
detail. And there's no coordinate conversion needed.