Re: Release check

[Jeremy Bettis]

> > > Under what circumstances would it be necessary to include Protocol.h?
> > > Does this affect just the libraries or also user code?

> Also, it seems like a pain to need to include Protocol.h to ask an object 
> if if conformsTo: @protocol(...), but this is the way NeXT did it so I 
> guess we are stuck with that.  If I were redesigning this from scratch I 
> would make sure all Objective-C language features were pulled in by one 
> include -- if that.

Protocol.h is only needed to send messages to a Protocol object.

I think that having NSObject.h not include Object.h is a great idea.  Also 
to make this discussion concrete, I have a sample program that represtents 
typical "user" use of Protocols.  This causes ZERO warnings in gcc 3.4.2

I put a NSObject inline here to show how it might work if NSObject did not 
include Object.h

/*
gcc -g -I$GNUSTEP_ROOT/System/Library/Headers -Wall prottest.m -o 
prottest.exe -L$GNUSTEP_ROOT/System/Library/Libraries -lgnustep-base -lobjc 
&& ./prottest
*/

#include <objc/objc.h>  // Defines id, BOOL and @class Protocol

@interface NSObject
+ (id) new;
+ (id) alloc;
- (id) init;
- (id) retain;
- (void) release;
- (void) dealloc;
- (BOOL) conformsToProtocol:(Protocol*)p;
@end

@protocol AA
- (void) foo;
@end

@interface A : NSObject <AA>
@end

@implementation A
- (void) foo
{

}
@end

int main()
{
A *a = [A new];
if ([a conformsToProtocol:@protocol(AA)]) {
 [a foo];
}
[a release];
return 0;
}


And here is a concrete example of unusual use of Protocols that might cause 
a warning in existing code if #include <objc/Protocol.h> was not there.

Protocol *aa = @protocol(AA);
[aa name];

Causes this warning:
warning: `Protocol' may not respond to `-name'
warning: (Messages without a matching method signature
warning: will be assumed to return `id' and accept
warning: `...' as arguments.)



I think we can all live with that.