drawrect Vs layoutsubviews - NSView - Cocoa touch

Question

I can't find a decent answer to explain me the difference between these 2 functions. when does every one gets called, and how does one different then the other ?

for example , can't I just layout my views inside drawrect ?

Thanks

Answer 1

-layoutSubviews is called from -layoutIfNeeded if the "layout needed" flag was set (using -setNeedsLayout, or automatically when the view bounds changes). Use it for positioning your view [EDIT: Use it for positioning subviews].

-drawRect: is called from -displayIfNeeded if the "display needed" flag was set (using -setNeedsDisplay, or automatically if you set view.contentMode = UIViewContentModeRedraw).

Both -layoutIfNeeded and -displayIfNeeded are called automatically by UIKit/CoreAnimation before things are drawn to screen; you rarely need to call them directly.

You can position your subviews in -drawRect: (you can even add subviews!), but this is unwise:

• By default, -setNeedsDisplay is not called automatically on a bounds change.
• Implementing -drawRect: reduces performance (UIKit/CoreAnimation has to create a bitmap-backed graphics context for you); only do so if you need to perform custom drawing.
• You need to redraw the view in -drawRect:. Drawing is expensive. Moving views around is cheap.
• UIKit/CoreAnimation probably does a layout pass followed by a drawing pass. CoreAnimation can use layout information to decide what views need drawing (e.g. it could ignore views obscured by opaque subviews, off-screen views, or subviews outside the bounds of a clipsToBounds=YES view; or it could only draw a sub-rect of a large view). If you move views during the drawing pass, CoreAnimation might not draw them correctly.

EDIT: And some more detail when I'm awake:

What's the difference between "display" and "draw"? Displaying is done by-[CALayer display]; the default implementation is (approximately)

• If the layer's delegate responds to -displayLayer:, call [self.delegate displayLayer:self]. -displayLayer: is supposed to set layer.content to (e.g.) a CGImage,
• Otherwise, if the layer's delegate responds to -drawLayer:inContext:, set up a bitmap-backed context, call [self.delegate drawLayer:self inContext:context], and save the output to layer.content (the output is actually a CABackingStore, which is presumably a private API)
• Otherwise, don't change layer.content.

The view is the layer's delegate, so you can implement -[MyView displayLayer:] instead, and do interesting stuff like

• self.layer.contents = (id)([UIImage imageNamed:@"foo"].CGImage) (which is roughly what UIImageView does)
• A no-op, to prevent any "drawing". This might be useful if you've subclassed e.g. UIToolbar and want to give it a transparent background. (This also prevents creation of the CGContext/CABackingStore.)
• Move subviews around without the performance penalty (but it's still not a good idea for the reasons above).