Returns a new self-description (aka metamodel).

You must never use this method to retrieve an entity description, but only retrieves it through a ETModelDescriptionRepository instance.

This method can be invoked at runtime by a repository to automatically collect the entity descriptions and make them available in this repository.

You can implement this method to describe your subclasses more precisely than -basicNewEntityDescription .

You must never call [super newEntityDescription] in the implementation.

You must not return an autoreleased object.

For example:

ETEntityDescription *desc = [self newBasicEntityDescription];

// For subclasses that don't override -newEntityDescription, we must not add the 
// property descriptions that we will inherit through the parent (the 
// 'MyClassName' entity description).
if ([[desc name] isEqual: [MyClass className]] == NO) return desc;

ETPropertyDescription *city = [ETPropertyDescription descriptionWithName: @"city" type: (id)@"NSString"];
ETPropertyDescription *country = [ETPropertyDescription descriptionWithName: @"country" type: (id)@"NSString"];

[desc setPropertyDescriptions: A(city, country)];

[desc setAbstract: YES];

return desc;

If you want set the parent explicitly, replace -newBasicEntityDescription with:

ETEntityDescription *desc = [ETEntityDescription descriptionWithName: [self className]];

// Will be resolved when the entity description is put in the repository
[desc setParent: NSStringFromClass([self superclass])];

Returns a new self-description (aka metamodel).

You must never use this method to retrieve an entity description, but only retrieves it through a ETModelDescriptionRepository instance.

This method can be invoked at runtime by a repository to automatically collect the entity descriptions and make them available in this repository.

You can implement this method to describe your subclasses more precisely than -basicNewEntityDescription .

You must never call [super newEntityDescription] in the implementation.

You must not return an autoreleased object.

For example:

ETEntityDescription *desc = [self newBasicEntityDescription];

// For subclasses that don't override -newEntityDescription, we must not add the 
// property descriptions that we will inherit through the parent (the 
// 'MyClassName' entity description).
if ([[desc name] isEqual: [MyClass className]] == NO) return desc;

ETPropertyDescription *city = [ETPropertyDescription descriptionWithName: @"city" type: (id)@"NSString"];
ETPropertyDescription *country = [ETPropertyDescription descriptionWithName: @"country" type: (id)@"NSString"];

[desc setPropertyDescriptions: A(city, country)];

[desc setAbstract: YES];

return desc;

If you want set the parent explicitly, replace -newBasicEntityDescription with:

ETEntityDescription *desc = [ETEntityDescription descriptionWithName: [self className]];

// Will be resolved when the entity description is put in the repository
[desc setParent: NSStringFromClass([self superclass])];

Returns a new minimal self-description without any property descriptions.

This entity description uses the class name as its name and the parent is set to the superclass name.

The parent will be resolved once when the description is added to the repository.

You must never use this method to retrieve an entity description, but only a ETModelDescriptionRepository instance to do so.

The returned object is not autoreleased.

See also -newEntityDescription .

Returns the description of the receiver by default.

Subclasses can override this method to return a string representation that encodes some basic infos about the receiver. This string representation can then be edited, validated by -validateValue:forKey:error: and used to instantiate another object by passing it to +objectWithStringValue: .

Returns YES when the receiver is an object which can be passed to -setObjectValue: or returned by -objectValue . Some common object values like string and number can be displayed and edited transparently (in an NSCell instance to take an example). If you define additional common object values, you usually have to write related formatters. Returns NO by default. Subclasses can override this method to specify an object can be accepted and used a common object value.

Returns YES if the receiver is an NSString instance, otherwise returns NO.

Returns YES if the receiver is an NSNumber instance, otherwise returns NO.

Returns the receiver description. Subclasses can override this method to return a more appropriate display name.

Returns the receiver description. Subclasses can override this method to return a more appropriate display name.

Returns the description as NSObject would. This method returns the same value as -description if the latter method isn't overriden in your subclasses, otherwise it returns the value that -description would return if you haven't overriden it. Useful to get consistent short descriptions on all instances and can be used to provide custom description built with other short descriptions.

Returns a description generated based on the given options.

Might describe a tree or graph structure if a traversal key is provided to recursively invoke -descriptionsWithOptions: on each object node. To do so, put ETDescriptionOptionTraversalKey with a valid KVC key in the options. You can also set a max depth with ETDescriptionOptionMaxDepth to limit the description size or end a graph traversal.

You can collect key path values on each object node by specifying an array of key paths with ETDescriptionOptionValuesForKeyPaths.

The description format is roughly: depth based indentation + object short description + keyPath1: value1, keyPath2: value2 etc.

By default, -description is used to print both object short description and key path values.

For customizing the object short description, put kETDescriptionOptionShortDescriptionSelector with a custom selector string in the options (-description is then used as fallback).

If you override -description to call -descriptionWithOptions: , you must provide a valid kETDescriptionOptionShortDescriptionSelector to prevent an endless loop (for example, just use -primitiveDescription).

For presenting each key path on a new line, put kETDescriptionOptionPropertyIndent with a tab string in the options.

Here is an example based on EtoileUI that dumps an item tree structure:

// ObjC code
ETLog(@"\n%@\n", [browserItem descriptionWithOptions: [NSMutableDictionary dictionaryWithObjectsAndKeys: 
	A(@"frame", @"autoresizingMask"), kETDescriptionOptionValuesForKeyPaths,
	@"items", kETDescriptionOptionTraversalKey, nil]]);

// Console Output
<ETLayoutItemGroup: 0x9e7b268> { frame: {x = 0; y = 0; width = 600; height = 300}, autoresizingMask: 18 }
	<ETLayoutItemGroup: 0x9fbea48> { frame: {x = 0; y = 0; width = 1150; height = 53}, autoresizingMask: 2 }
		<ETLayoutItem: 0x9f29240> { frame: {x = 12; y = 12; width = 100; height = 22}, autoresizingMask: 0 }
		<ETLayoutItem: 0x9e6fcf0> { frame: {x = 124; y = 12; width = 100; height = 24}, autoresizingMask: 0 }
	<ETLayoutItemGroup: 0x9fac170> { frame: {x = 0; y = 0; width = 1150; height = 482}, autoresizingMask: 18 }
		<ETLayoutItemGroup: 0x9fb2870> { frame: {x = 0; y = 0; width = 50; height = 50}, autoresizingMask: 0 }

options must not be nil, otherwise raises an NSInvalidArgumentException.

You can override this method in subclasses, although it is not advised to. The options dictionary can be changed arbitrarily in a new implementation.

Returns an empty set.

Overrides to return the receiver key paths to be observed when an observer is set up with -addObserver: .

The returned set content must not change during the whole object lifetime, otherwise -removeObserver: will crash randomly.

Returns an empty set.

Overrides to return the receiver key paths to be observed when an observer is set up with -addObserver: .

The returned set content must not change during the whole object lifetime, otherwise -removeObserver: will crash randomly.

Sets up the given object to observe each receiver key paths returned by -observableKeyPaths .

The observer will receive NSKeyValueObservingOptionOld and NSKeyValueObservingOptionNew in the change dictionary.

Removes the observer that was observing the receiver key paths returned by -observableKeyPaths .

Returns a mutable counterpart class or Nil if such a class does not exist.

Returns a mutable counterpart class or Nil if such a class does not exist.

Returns YES if the receiver is declared as mutable, otherwise returns NO.

This method returns NO by default. You can override it to return YES if you want to declare your subclass instances as mutable objects (which are collections most of time).

If you adopts ETCollectionMutation in a subclass, you don't need to override this method to declare your collection objects as mutable.

Returns YES if the receiver is declared as a collection by conforming to ETCollection protocol, otherwise returns NO.

You must never override this method in your collection classes, you only need to adopt ETCollection protocol.

Returns YES if the receiver is declared as a collection by conforming to ETCollectionMutation protocol, otherwise returns NO.

You must never override this method in your collection classes, you only need to adopt ETCollectionMutation protocol.

Returns YES if the receiver is a low-level collection such as NSArray, NSet, etc., otherwise returns NO.

For a model object such as ETLayoutItemGroup that conforms to ETCollection protocol, would return NO.

Returns YES if the receiver is declared as a group, otherwise returns NO.

This method returns NO by default. You can override it to return YES if you want to declare your subclass instances as groups.

A group is specialized model object which is a composite and can behave like a mutable collection. A basic collection object (like NSMutableArray, NSMutableDictionary, NSMutableSet) must never be declared as a group.

COGroup in CoreObject or ETLayoutItemGroup in EtoileUI are typical examples.

A group should conform to ETCollectionMutation protocol.

Returns a key for inserting the receiver into the given keyed collection.

By default, returns a key built by incrementing the integer value in the 'Unknown <number>' pattern, until it provides a key not yet in use in the collection argument.

This key is retrieved by a collection in reply to -insertObjects:atIndexes:hints: of ETCollectionMutation protocol. You can return different keys depending on the type of collection. This parameter is usually the mutated collection itself.

Can be overriden to return YES in order to support exposing properties, in case -valueForProperty: and -setValue:forProperty: access another object and not the receiver.

This method is used by ETMutableObjectViewpoint to determine whether the receiver represented property can be accessed through -valueForProperty: or -valueForKey: .

By default, returns NO.

If -valueForProperty: and -setValue:forProperty: don't access a represented object, there is not need to override this method to return YES in subclasses. You usually never need to override this method unless you adopt ETPropertyViewpoint protocol.

Returns the names of the properties exposed by the receiver through -valueForProperty: and -setValue:forProperty: , or -valueForKey: and -setValue:forKey: if -requiresKeyValueCodingForAccessingProperties returns YES.

Returns both the property names bound to the object entity description and the basic property names.

+[ETModelDescriptionRepository mainRepository] is used to look up the entity description.

To be exposed through Property Value Coding, the receiver properties must be listed among the returned properties.

Can be overriden to return property names bound to entity descriptions that don't belong to the main repository, or filter some properties out. In the overriden method, you should usually return -basicPropertyNames along the property description names.

For a NSObject subclass not bound to an entity description, the property names related to the closest superclass bound to an entity description are returned through a recursive lookup in -entityDescriptionForClass: .

See -basicPropertyNames , -valueForProperty: and -setValue:forProperty: . See also -[ETPropertyValueCoding propertyNames] .

Returns the value of the property.

If the property doesn't exist, returns nil but must not raise an exception.

The method precise semantic is under the control of the class implementing the protocol. For example, the property can belong to another object than the receiver.

See also -[NSObject valueForProperty:] and -[ETPropertyViewpoint valueForProperty:] .

Sets the value of the property and returns YES if the value was successfully set.

If the property doesn't exist, returns NO but must not raise an exception.

The method precise semantic is under the control of the class implementing the protocol.

See also -[NSObject setValue:forProperty:] and -[ETPropertyViewpoint setValue:forProperty:] .

Returns whether the receiver is a ETKeyValuePair instance.

Returns the method name prefix used to build the method name to be invoked by the double dispatch.

By default, returns visit.

Can be overriden to return a custom prefix such as visit.

See also -visit: .

Tries to build a method name based on the given object type and invoke it.

For example, if you have a NSView, and you pass it as an argument to -visit: , the selector visitNSView: is built and invoked with the given view on the receiver. If the receiver doesn't respond to visitNSView: , then visitView: is built by trimming the class name prefix, and invoked. If the receiver still doesn't respond the last built selector, then it fails silently and returns nil.

Class name prefix are trimmed based on the value returned by +[NSObject(Etoile) typePrefix] . You can override this last method to return a custom prefix, by default it returns NS.

If you want to use another method name prefix than visit (e.g. to build a selector such as renderView:), -doubleDispatchPrefix can be overriden.

Subclasses can override this method, if they want to customize the double-dispatch behavior.

Does the same than -visit: but reports whether a double-dispatch method was succesfully invoked by setting performed to YES, or NO when no such method was found.

See also -visit: and -supportsDoubleDispatchWithObject:.

This method is called by -visit: and implements the double-dispatch logic.

Returns whether the receiver implements a double dispatch method that corresponds to the given object type.

This method serves a similar purpose than -[NSObject respondsToSelector:] .

See also -visit: .

Builds and returns the selector to be invoked for a double dispatch on the given type.

For a detailed example, see -visit: .

Can be overriden in subclasses to implement an alternative strategy to build the method names targeted by the double dispatch.

Tries to invoke the selector with the given object as first argument, and returns either the value returned by the invoked method or nil.

If the receiver doesn't respond to the selector, performed is set to NO and nil is returned, otherwise performed is set YES.

Apply aClass to this class as a trait.

Raises exceptions if the trait application cannot be validated.

Apply aClass to this class as a trait, without the trait methods listed in excludedNames, and by renaming the trait methods with the name values provided in the aliasedNames dictionary (where keys should be the existing trait method names).

Raises exceptions if the trait application cannot be validated.

Does the same than +applyTraitFromClass:excludedMethodNames:aliasedMethodNames: but allows to replace methods in the target class with trait methods if YES is passed as the last argument.

By default, the trait overriding rule states that trait methods cannot replace methods that belongs to the target class, but only hide methods declared in superclasses and inherited by the target class.

If YES is passed and there are other traits applied to the target class, the ordering use to apply traits cannot be ignored anymore.

Raises exceptions if the trait application cannot be validated.

Returns all descendant subclasses of the receiver class.

The returned array doesn't include the receiver class.

You should rather use -[ETClassMirror allSubclassMirrors] , this method could be deprecated in the future.

Returns all subclasses which inherit directly from the receiver class.

Subclasses that belongs to the class hierarchy of the receiver class but whose superclasses aren't equal to it, are excluded.

The returned array doesn't include the receiver class.

Returns the uniform type identifier of the object.

The UTI object encodes the type of the object in term of namespaces and multiple inheritance.

By default, the UTI object is shared by all instances by being built from the class name. If you need to introduce type at instance level, you can do it by overriding this method.

Returns the type name which is the last component of the string value returned by the receiver UTI object.

See also -UTI.

Returns the type prefix, usually the prefix part of the type name returned by -className .

By default, returns 'NS'.

You must override this method in your subclass to indicate the prefix of your new class name. Take note the prefix will logically apply to every subsequent subclasses inheriting from the class that redefines -typePrefix .

Returns the class name.

This Foundation method is missing on iOS.

Handle a new connection request received by listenerSocket.

Handle data received over the specified socket.

Inserts the given collection elements at separate indexes.

When the collection is not ordered, the indexes are ignored.

The element are inserted one-by-one by increasing index value while iterating over the indexes. When the greatest index is reached and several elements remain to be inserted, they are inserted at that same index.

For a more precise description of the behavior ordered collection should comply to, see -[NSArray insertObjects:atIndexes:] in Cocoa documentation.

Removes the elements from the collection.

Removes the elements at the given indexes from the collection.

You should only implement this method when the collection is ordered.

Tells the receiver that the collection bound to the property is about to change.

By default, limited to calling:

-willChangeValueForKey:

NSDictionary, NSMapTable and NSIndexSet

-willChange:valuesAtIndexes:forKey:

NSArray, NSPointerArray and NSOrderedSet

-willChangeValueForKey:withSetMutation:usingObjects:

NSSet and NSHashTable

For an ordered collection, the correct index must be passed for each inserted or removed object, otherwise a NSInvalidArgumentException is raised.

For an unordered collection, the indexes should be empty.

For nil arguments, raises a NSInvalidArgumentException.

Can be overriden, but the superclass implementation must be called.

Tells the receiver that the collection bound to the property is about to change.

By default, limited to calling:

-didChangeValueForKey:

NSDictionary, NSMapTable and NSIndexSet

-didChange:valuesAtIndexes:forKey:

NSArray, NSPointerArray and NSOrderedSet

-didChangeValueForKey:withSetMutation:usingObjects:

NSSet and NSHashTable

For an ordered collection, the correct index must be passed for each inserted or removed object, otherwise a NSInvalidArgumentException is raised.

For an unordered collection, the indexes should be empty.

For nil arguments, raises a NSInvalidArgumentException.

Can be overriden, but the superclass implementation must be called.

Returns whether the receiver is an index-value pair. By default, returns NO. See also -[ETIndexValuePair isIndexValuePair] .

Adds an instance method to the class, using the specified block. The types for the method are inferred from the block.

The block must take an object (self) as the first argument. Any subsequent arguments are the explicit arguments to the method. The _cmd argument is not available for methods added in this way.

Adds a class method to the class, using the specified block. The types for the method are inferred from the block.

The block must take an object (self) as the first argument. Any subsequent arguments are the explicit arguments to the method. The _cmd argument is not available for methods added in this way.

Adds a method to the receiver, using the specified block. The types for the method are inferred from the block.

The block must take an object (self) as the first argument. Any subsequent arguments are the explicit arguments to the method. The _cmd argument is not available for methods added in this way.

Adds the specified method to this instance. Objects modified in this way get a hidden dictionary for non-indexed instance variables, allowing them to use KVC to set arbitrary objects on self.

Returns a clone of the object. The clone will inherit all methods and associated objects. To copy instance variables, you must override this method.

Returns YES if this object inherits from another object.

Returns the prototype for this object, or nil if this object does not have one.

Does the same as valueForKey:, except when this object is a prototype and a block closure is associated with the supplied key. The block closure is then returned without being invoked.

Returns the receiver itself when it can respond to the next message that follows -ifResponds , otherwise returns nil.

If we suppose the Cat class doesn't implement -bark , then -ifResponds would return nil and thereby -bark be discarded: [[cat ifResponds] bark];

Now let's say the Dog class implement -bark , the -ifResponds will return 'dog' and -bark be executed: [[dog ifResponds] bark];

Returns NO.

Returns the array representation used to evaluate high-order messages.

If not implemented, -contentArray is used.

If info is not NULL, you can return additional information to be passed to the element-handling hook -placeObject:inCollection:insteadOfObject:atIndex:havingAlreadyMapped:info:.

If the receiver conforms to the ETCollection protocol, this method returns a proxy that will let -map and -filter methods iterate over the contents of the collection when it is used as an argument to a message. This way, [[people map] sendMail: [messages each]]; will cause the -sendMail: message to be executed once with every combination of elements from the people and the messages collection.

Note 1: It is only possible to use an proxy object created with -each if it is passed as an argument to a message that is send to a higher-order messaging proxy created by -filter , -map, -filteredCollection or -mappedCollection . Doing [aCollection addObject: [things each]] won't do anything.

Note 2: If an each proxy is passed to a message used as a filter predicate, it suffices that the predicate evaluates to YES for one element of the argument-collection. If a collection A contains "foo" and "bar" and collection B contains "BAR" and "bar", after [[A filter] isEqualToString: [B each]]; , A will still contain "bar" (but not "BAR"), since one of the elements of B matched "bar".

Records the call stack symbols with the shared stack trace recorder.

Returns an array of stack traces previously recorded with the shared stack trace recorded for the receiver.