Is there any way to discover at runtime which subclasses exist of a given class?
Edit: From the answers so far I think I need to clarify a bit more what I am trying to do. I am aware that this is not a common practice in Cocoa, and that it may come with some caveats.
I am writing a parser using the dynamic creation pattern. (See the book Cocoa Design Patterns by Buck and Yacktman, chapter 5.) Basically, the parser instance processes a stack, and instantiates objects that know how to perform certain calculations.
If I can get all the subclasses of the MYCommand class, I can, for example, provide the user with a list of available commands. Also, in the example from chapter 5, the parser has an substitution dictionary so operators like +, -, * and / can be used. (They are mapped to MYAddCommand, etc.) To me it seemed this information belonged in the MyCommand subclass, not the parser instance as it kinda defeats the idea of dynamic creation.
Not directly, no. You can however get a list of all classes registered with the runtime as well as query those classes for their direct superclass. Keep in mind that this doesn't allow you to find all ancestors for the class up the inheritance tree, just the immediate superclass.
You can use objc_getClassList() to get the list of Class objects registered with the runtime. Then you can loop over that array and call [NSObject superclass] on those Class objects to get their superclass' Class object. If for some reason your classes do not use NSObject as their root class, you can use class_getSuperclass() instead.
I should mention as well that you might be thinking about your application's design incorrectly if you feel it is necessary to do this kind of discovery. Most likely there is another, more conventional way to do what you are trying to accomplish that doesn't involve introspecting on the Objective-C runtime.
Rather than try to automatically register all the subclasses of MYCommand, why not split the problem in two?
First, provide API for registering a class, something like +[MYCommand registerClass:].
Then, create code in MYCommand that means any subclasses will automatically register themselves. Something like:
#implementation MYCommand
+ (void)load
{
[MYCommand registerClass:self];
}
#end
Marc and bbum hit it on the money. This is usually not a good idea.
However, we have code on our CocoaHeads wiki that does this: http://cocoaheads.byu.edu/wiki/getting-all-subclasses
Another approach was just published by Matt Gallagher on his blog.
There's code in my runtime browser project here that includes a -subclassNamesForClass: method. See the RuntimeReporter.[hm] files.
Related
Fairly early on in my app, when I was a lot less experienced than I am now, I wanted to spice up some transitions between view controllers with my own custom animations. Having no idea where to start, I looked around SO for a pattern like MVC that could be accessed from nearly any controller at any time, and as it turns out, a singleton was the way to go.
What I didn't realize is that there seems to be a strong and well-defended hatred of the singleton pattern, and I myself am starting to see why, but that is beside the point.
So, a while later, I decided to move my very same implementation into a category on UINavigationController (after all, it handles transitions!), kept the original classes around for comparison, and am wondering which method would work best. Having thoroughly tested both implementations, I can say without a doubt that they are equal in every way, including speed, accuracy, smoothness, frame-rate, memory usage, etc. so which one is 'better' in the sense of overall maintainability?
EDIT: after reading the well-written arguments you all have made, I have decided to use a singleton. #JustinXXVII has made the most convincing argument (IMHO), although I consider every answer here equally worthy of merit. Thank you all for your opinions, I have upvoted all answers in the question.
I believe the best option is use the category.
Because if you are already using UINavigationController, do not make sense create a new class that will only manage the transition, like you told: (after all, it handles transitions!)
This will be a better option to maintain your code, and you will be sure that the thing do what they expect to do, and if you already have an instance that do the transitions, why create another?
The design patterns, like singleton, factory, and others, need to be used with responsibility. In your case, I do not see why use a singleton, you use it only to no instantiate new objects, you do not really need to have only one instance of it, but you do it because you want only one.
I'll make the case for a singleton object. Singletons are used all over UIKit and iOS. One thing you can't do with categories is add instance variables. There are two things about this:
MVC workflows don't tolerate objects with intimate knowledge of other objects
Sometimes you just need a place to reference an object that doesn't really belong anywhere else
These things go against each other, but the added ability to be able to keep an instance variable that doesn't really have an "owner" is why I favor the singleton.
I usually have one singleton class in all of my XCode projects, which is used to store "global" objects and do mundane things that I don't want to burden my AppDelegate with.
An example would be serializing/archiving objects and unarchiving/restoring. I have to use the same method throughout several classes, I don't want to extend UIViewController with some serializing method to write and read arbitrary files. Maybe it's just my personal preference.
I also might need a quick way to lookup information in NSUserDefaults but not want to always be writing [[NSUserDefaults standardUserDefaults]stringForKey:#"blah"], so I will just declare a method in my singleton that takes a string argument.
Until now i've not really thought too much about using a category for these things. One thing is sure though, I'd rather not be instantiating a new object a hundred times to do the same task when I can have just one living object that sticks around and will take care of stuff for me. (Without burdening the AppDelegate)
I think that the real question is in "design" (as you said, both codes work fine), and by writing down your problem in simple sentences, you will find your answer :
singleton's purpose is to have only one instance of a class running in your app. So you can share things between objects. (one available to many objects)
category purpose is to extend the methods available to a class. (available to one class of objects only ! ok...objects from subclasses too)
what you really want is to make a new transition available to UINavigationController class. UINavigationController, which has already some method available to change view (present modal views, addsubviews, etc.) is built to manage views with transitions (you said it yourself, it handles transitions), all you want to do is adding another way of handling transitions for your navigation controllers thus you would preferably use a category.
My opinion is that what you want to achieve is covered by the category and by doing this you ensure that the only objects which are accessing this method are entitled to use it. With the singleton pattern, any object of any class could call your singleton and its methods (and... it could work nobody knowing how for an OS version n but your app could be broken in n+1 version).
In this implementation, for which there is no need to use a Singleton, there may be no difference at all. That doesn't mean that there isn't one.
A plastic bucket holds as much water as a metal bucket does, and it does it just as well. In that aspect there seems to be no difference between the two. However, if you try to transport something extremely hot, the plastic bucket might not do the job so well..
What I'm trying to say is, they both serve their purposes but in your case there seemed to be no difference because the task was too generic. You wanted a method that was available from multiple classes, and both solutions can do that.
In your case, however, it might be a whole of a lot simpler to use a Category. The implementation is easier and you (possibly) need less code.
But if you were to create a data manager that holds an array of objects that you ONLY want available at one place, a Category will not be up to the task. That's a typical Singleton task.
Singeltons are single-instance objects (and if made static, available from nearly everywhere). Categories are extensions to your existing classes and limited to the class it extends.
To answer your question; choose a Category.
*A subclass might also work, but has its own pros and cons
Why don't you simply create a base UIViewController subclass and extend all of your view controllers from this object? A category doesn't make sense for this purpose.
Singletons, as the name suggests, has to be used when there is a need to be exactly one object in your application. The pattern for the accessor method ensures only this requirement being a class method:
+ (MyClass*) sharedInstance
{
static MyClass *instance = nil;
if (instance == nil) instance = [[MyClass alloc] init];
return instance;
}
If implemented well, the class also ensures that its constructor is private thus nobody else can instantiate the class but the accessor method: this ensures that at any time at most one instance of the class exists. The best example of such class is UIApplication since at any time there might be only one object of this class.
The point here is that this is the only requirement towards singleton. The role of the accessor method is to ensure that there is only one instance, and not that it would provide access to that instance from everywhere. It is only a side effect of the pattern that, the accessor method being static, everybody can access this single object without having a reference (pointer) to it a priori. Unfortunately this fact is widely abused by Objective C programmers and this leads to messed up design and the hatred towards singleton pattern you mentioned. But all in all it is not the fault the singleton patter but the misuse of their accessor method.
Now turning back to your question: if you don't need static / global variables in your custom transition code (I guess you don't) then the answer is definitely go for categories. In C++ you would subclass from some parent BaseTransition class and implement your actual drawing methods. Objective C has categories (that in my opinion is another way that easily messes up the design, but they are much more convenient) where you can add custom functionality even accessing the variables of your host class. Use them whenever you can redeem singletons with them and don't use singletons when the main requirement towards your class is not that it would be only one instance of it.
I wonder why smalltalk doesn't make use of java-style inner class. This mechanism effectively allows you to define a new instance of a new class, on-the-fly, where you need it, when you need it. It comes handy when you need an object conforming to some specific protocol but you don't want to create a normal class for it, because of its temporary and local nature being very implementation specific.
As far I know, it could be done easily, since syntax for subclassing is standard message sending. And you can pass self to it so it has the notion of the "outer" object. The only issue is anonymousity - the class should not be present in object browser and must be garbage collected when no instances of it exit.
The question is: Has anyone thought of this?
There are really two answers here:
1 - Yes, it is not hard to create anonymous classes that automatically get garbage collected. In Squeak they are called "uniclasses" because the typical use case is for adding methods to a single object. Systems that use this are for example Etoys and Tweak (although in Etoys the classes are actually put into the SystemDict for historic reasons). Here's some Squeak code I recently used for it:
newClass := ClassBuilder new
newSubclassOf: baseClass
type: baseClass typeOfClass
instanceVariables: instVars
from: nil.
baseClass removeSubclass: newClass.
^newClass
Typically, you would add a convenience method to do this. You can can then add methods, and create an instance, and when all instances are gone, the class will be gc'ed too.
Note that in Java, the class object is not gc'ed - an inner class is compiled exactly like a regular class, it's only hidden by the compiler. In contrast, in Smalltalk this all happens at runtime, even the compiling of new methods for this class, which makes it comparatively inefficient. There is a much better way to create anonymous precompiled behavior, which brings us to answer 2:
2 - Even though it's not hard, it's rarely used in Smalltalk. The reason for that is that Smalltalk has a much more convenient mechanism. Inner classes in Java are most often used for making up a method on the fly implementing a specific interface. The inner class declaration is only needed to make the compiler happy for type safety. In Smalltalk, you simply use block closures. This lets you create behavior on the fly that you can pass around. The system libraries are structured in a way to make use of block closures.
I personally never felt that inner classes were something Smalltalk needed.
If you are thinking of using inner classes for tests, then you can also take a look to the class ClassFactoryForTestCase
Creating an anonymous class(es) in smalltalk is a piece of cake.
More than that, any object which has 3 its instance variables properly set to: its superclass, method dictionary and instance format could serve as a class (have instances).
So, i don't see how the problem here.
If you talking about tool(s) support, like browsing or navigating code which contained in such classes, this is different story. Because by default all classes in system are public, and system dictionary is a flat namespace of them (yes , some implementations has namespaces). This simple model works quite well most of the times.
I am pretty sure it could be done with some hacking around the Class and Metaclass protocol. And the question pops quite often from people who have more experience in Java, and Smalltalk becomes interesting to them. Since inner classes have not been implemented inspite of that, I take it to be the sign that most Smalltalk users do not find them usable. This might be because Smalltalk has blocks, which in simpler manner solve many if not all problems that led to the introduction of inner classes to Java.
(a) You could send the messages to create a new class from inside the method of another class
(b) I doubt that there is any benefit in hiding the resulting class from the introspection system
(c) The reason you use inner classes in Java is because there are no first-class functions. If you need to pass a piece of code in Smalltalk, you just pass a block. You don't need to wrap it up with some other type of object to do so.
The problem (in Squeak at least) comes from the lack of a clean separation of concerns. It's trivial to create your own subclass and put it in a private SystemDictionary:
myEnv := SystemDictionary new.
myClass := ClassBuilder new
name: 'MyClass'
inEnvironment: myEnv
subclassOf: Object
type: #normal
instanceVariableNames: ''
classVariableNames: ''
poolDictionaries: ''
category: 'MyCategory'
unsafe: false.
But even though you put that class in your own SystemDictionary, the 'MyCategory' category added to the system navigation (verifiable by opening a Browser), and - worse - the class organisers aren't created, so when you navigate to MyClass you get a nil pointer.
It's certainly not impossible, theoretically. Right now the tooling's geared towards a single pool of globally visible class definitions.
I've recently discovered categories and was wondering when it might be appropriate to use them in a user defined class/new class. For example, I can see the benefits of adding a category to an existing class like NSString, but when creating a new class what would be the advantage of adding a category to this rather than just implementing a normal method?
Hope this makes sense.
Many thanks
Jules
The answer isn't really any different for your own classes than it is for framework classes. If you have multiple projects, you'll likely end up sharing some classes between them. However, you may want to extend some of your classes so that they work more easily with a specific project, but not want to include those extra methods in your other projects, where they might not make sense. You can use a category to extend your class without needing to subclass.
If I understand your question correctly, creating a "new class" is always "subclassing" because you're subclassing NSObject at the very least.
You could use categories on a new class to separate out sections of responsibility of a complex class. For example, all the basic functionality (instance variables, accessors, description, etc.) can go in one file (the "main" class file) while all methods to support a protocol (such as NSTableViewDataSource) can go in another.
Some take this approach to keep things "neat". I'm a firm believer in "if it's my own custom class, all its code should be in one file" so I do not personally do this. I demarcate different logical aspects of the class' code with "#pragma mark Some Section Name" to help navigation and readability. Your mileage may vary.
Adding a Category on NSString is useful when you want to call a method on every single NSString instance you will encounter. This is a real improvement over inheritance for this kind of object because they are used by the core framework and you don't have to convert a NSString object to your subclass when you want to call your custom method.
On the other hand, you can just put methods in, no instance variables.
In the book Refactoring by Martin Fowler, he has a section titled "Introduce Foreign Method" (A server class you are using needs an additional method, but you can't modify the class.) That's what categories are good for.
That said, there are times when using a category, instead of changing the class, is appropriate. A good example on using a category, even though you could change the server class, is how Apple handled the UIViewController MediaPlayer Additions. They could have put these two methods in UIViewController itself but since the only people who would ever use them are people who are using the Media Player framework, it made more sense to keep the methods there.
I had a bunch of objects which were responsible for their own construction (get properties from network message, then build). By construction I mean setting frame sizes, colours, that sort of thing, not literal object construction.
The code got really bloated and messy when I started adding conditions to control the building algorithm, so I decided to separate the algorithm to into a "Builder" class, which essentially gets the properties of the object, works out what needs to be done and then applies the changes to the object.
The advantage to having the builder algorithm separate is that I can wrap/decorate it, or override it completely. The object itself doesn't need to worry about how it is built, it just creates a builder and 'decorates' the builder with extra the functionality that it needs to get the job done.
I am quite happy with this approach except for one thing... Because my Builder does not inherit from the object itself (object is large and I want run-time customisation), I have to expose a lot of internal properties of the object.
It's like employing a builder to rebuild your house. He isn't a house himself but he needs access to the internal details, he can't do anything by looking through the windows. I don't want to open my house up to everyone, just the builder.
I know objects are supposed to look after themselves, and in an ideal world my object (house) would build itself, but I am refactoring the build portion of this object only, and I need a way to apply building algorithms dynamically, and I hate opening up my objects with getters and setters just for the sake of the Builder.
I should mention I'm working in Obj-C++ so lack friend classes or internal classes. If the explanation was too abstract I'd be happy to clarify with something a little more concrete. Mostly just looking for ideas or advice about what to do in this kind of situation.
Cheers folks,
Sam
EDIT: is it a good approach to declare a
interface House(StuffTheBuilderNeedsAccessTo)
category inside Builder.h ? That way I suppose I could declare the properties the builder needs and put synthesizers inside House.mm. Nobody would have access to the properties unless they included the Builder header....
That's all I can think of!
I would suggest using Factory pattern to build the object.
You can search for "Factory" on SO and you'll a get a no. of questions related to it.
Also see the Builder pattern.
You might want to consider using a delegate. Add a delegate method (and a protocol for the supported methods) to your class. The objects of the Builder class can be used as delegates.
The delegate can implement methods like calculateFrameSize (which returns a frame size) etc. The returned value of the delegate can be stored as an ivar. This way the implementation details of your class remain hidden. You are just outsourcing part the logic.
There is in fact a design pattern called, suitable enough, Builder which does tries to solve the problem with creating different configurations for a certain class. Check that out. Maybe it can give you some ideas?
But the underlying problem is still there; the builder needs to have access to the properties of the object it is building.
I don't know Obj-C++, so I don't know if this is possible, but this sounds like a problem for Categories. Expose only the necessary methods to your house in the declaration of the house itself, create a category that contains all the private methods you want to keep hidden.
What about the other way around, using multiple inheritance, so your class is also a Builder? That would mean that the bulk of the algorithms could be in the base class, and be extended to fit the neads of you specific House. It is not very beautiful, but it should let you abstract most of the functionality.
Can some one explain to me the difference between categories and inheritance in Objective C? I've read the entry in Wikipedia and the discussion on categories there doesn't look any different to that of inheritance. I also looked at the discussion on the topic in the book "Open iPhone Development" and I still don't get it.
Sometimes, inheritance just seems like more trouble than it is worth. It is correctly used when you want to add something to an existing class that is a change in the behaviour of that class.
With a Category, you just want the existing object to do a little more. As already given, if you just want to have a string class that handles compression, you don't need to subclass the string class, you just create a category that handles the compression. That way, you don't need to change the type of the string classes that you already use.
The clue is in the restriction that categories only add methods, you can't add variables to a class using categories. If the class needs more properties, then it has to be subclassed.(edit: you can use associative storage, I believe).
Categories are a nice way to add functionality while at the same time conforming to an object oriented principle to prefer composition over inheritance.
Edit January 2012
Things have changed now. With the current LLVM compiler, and the modern, 64-bit runtime, you can add iVars and properties to class extensions (not categories). This lets you keep private iVars out of the public interface. But, if you declare properties for the iVars, they can still be accessed / changed via KVC, because there is still no such thing as a private method in Objective-C.
Categories allow you to add methods to existing classes. So rather than subclass NSData to add your funky new encryption methods, you can add them directly to the NSData class. Every NSData object in your app now has access to those methods.
To see how useful this can be, look at: CocoaDev
One of favorite illustrations of Objective-c categories in action is NSString. NSString is defined in the Foundation framework, which has no notion of views or windows. However, if you use an NSString in a Cocoa application you'll notice it responds to messages like – drawInRect:withAttributes:.
AppKit defines a category for NSString that provides additional drawing methods. The category allows new methods to be added to an existing class, so we're still just dealing with NSStrings. If AppKit instead implemented drawing by subclassing we'd have to deal with 'AppKitStrings' or 'NSSDrawableStrings' or something like that.
Categories let you add application or domain specific methods to existing classes. It can be quite powerful and convenient.
If you as a programmer are given a complete set of source code for a code library or application, you can go nuts and change whatever you need to achieve your programming goal with that code.
Unfortunately, this is not always the case or even desirable. A lot of times you are given a binary library/object kit and a set of headers to make do with.
Then a new functionality is needed for a class so you could do a couple of things:
create a new class whole instead of a stock class -- replicating all its functions and members then rewrite all the code to use the new class.
create a new wrapper class that contains the stock class as a member (compositing) and rewrite the codebase to utilize the new class.
binary patches of the library to change the code (good luck)
force the compiler to see your new class as the old one and hope it does not depend on a certain size or place in memory and specific entry points.
subclass specialization -- create subclasses to add functionality and modify driver code to use the subclass instead -- theoretically there should be few problems and if you need to add data members it is necessary, but the memory footprint will be different. You have the advantage of having both the new code and the old code available in the subclass and choosing which to use, the base class method or the overridden method.
modify the necessary objc class with a category definition containing methods to do what you want and/or override the old methods in the stock classes.
This can also fix errors in the library or customize methods for new hardware devices or whatever. It is not a panacea, but it allows for class method adding without recompiling the class/library that is unchanged. The original class is the same in code, memory size, and entry points, so legacy apps don't break. The compiler simply puts the new method(s) into the runtime as belonging to that class, and overrides methods with the same signature as in the original code.
an example:
You have a class Bing that outputs to a terminal, but not to a serial port, and now that is what you need. (for some reason). You have Bing.h and libBing.so, but not Bing.m in your kit.
The Bing class does all kinds of stuff internally, you don't even know all what, you just have the public api in the header.
You are smart, so you create a (SerialOutput) category for the Bing class.
[Bing_SerialOutput.m]
#interface Bing (SerialOutput) // a category
- (void)ToSerial: (SerialPort*) port ;
#end
#implementation Bing (SerialOutput)
- (void)ToSerial: (SerialPort*) port
{
... /// serial output code ///
}
#end
The compiler obliges to create an object that can be linked in with your app and the runtime now knows that Bing responds to #selector(ToSerial:) and you can use it as if the Bing class was built with that method. You cannot add data members only methods and this was not intended to create giant tumors of code attached to base classes but it does have its advantages over strictly typed languages.
I think some of these answers at least point to the idea that inheritance is a heavier way of adding functionality to an existing class, while categories are more lightweight.
Inheritance is used when you're creating a new class hierarchy (all the bells and whistles) and arguably brings alot of work when chosen as the method of adding functionality to existing classes.
As someone else here put it... If you are using inheritance to add a new method for example to NSString, you have to go and change the type you're using in any other code where you want to use this new method. If, however, you use categories, you can simply call the method on existing NSString types, without subclassing.
The same ends can be achieved with either, but categories seem to give us an option that is simpler and requires less maintenance (probably).
Anyone know if there are situations where categories are absolutely necessary?
A Category is like a mixin: a module in Ruby, or somewhat like an interface in Java. You can think of it as "naked methods". When you add a Category, you're adding methods to the class. The Wikipedia article has good stuff.
The best way to look at this difference is that:
1. inheritance : when want to turn it exactly in your way.
example : AsyncImageView to implement lazy loading. Which is done by inheriting UIView.
2. category : Just want to add a extra flavor to it.
example : We want to replace all spaces from a textfield's text
#interface UITextField(setText)
- (NSString *)replaceEscape;
#end
#implementation UITextField(setText)
- (NSString *)replaceEscape
{
self.text=[self.text stringByTrimmingCharactersInSet:
[NSCharacterSet whitespaceCharacterSet]];
return self.text;
}
#end
--- It will add a new property to textfield for you to escape all white spaces. Just like adding a new dimension to it without completely changing its way.