Re: NSWindow frame handling (Was: [Gnustep-cvs] Commit Update)

[Alexander Malmberg]

Fred Kiefer wrote:
> Hi Alex,
>
> this changes look great, as they clean up an area that has been obscure 
> for a long time. Two comments though, the new methods on NSWindow should 
> be flagged with the #ifndef NO_GNUSTEP #endif preprocessor statements to 
> make clear that these are GNUstep extensions. As our headers will be 
> used by different people, some sticking only to the old OpenStep 
> specification, some using the current Cocoa interfaces, we should at 
> least hint, which methods are portable and which are GNUstep specific.

True. I'll fix this soon.

> And the other thing that wonders me is, if the variable windowDecorator 
> in NSWindow.m shouldn't be of type CLASS? You made it 
> id<GSWindowDecorator> which is, as far as I understand it, an object 
> which repsonds to the protocol GSWindowDecorator, but acutally we have a 
> class object here, of a subclass of GSWindowDecorationView.

Sure, but that's ok. Class objects are objects too, and the code just 
wants an object (any object, thus id) that conforms to the protocol.

> Not that
> this causes to much differences, but isn't there a better way in 
> Objective-C to express this?

A "class that conforms to a protocol"-type can't (currently) be 
expressed in objective-c.

- Alexander Malmberg