Nested #import statements: how to hide nested ones? - objective-c

I have the following project structure:
Main.h
Son1.h
Son2.h
They're not related (no father/son relationships), just two imports, like in Main.h I have:
#import Son1.h
#import Son2.h
If, from another file I write
#import Main.h
I will see all the methods/properties of Main.h (and that's ok) but I will also see the methods of Son1.h and Son2.h.
How can I prevent this?

It's important to understand what #import (and the C version: #include) do.
Compilation of C programs conceptually have 3 steps:
Preprocessing
Compiling
Linking
The directives beginning with # are acted on by the preprocessor (it doesn't matter at all whether this is a separate binary or part of the compiler). Both of these statements cause the contents of the named file to inserted into the importing (or including) file at the point where the directive is found. This is completely language-agnostic. The imported file can have anything in it. You can import a JPG, if you want (but good luck compiling!).
In your example, by importing Son1.h and Son2.h in Main.h, you are creating, from the compiler's perspective, a single file with the contents of Son1.h followed by the contents of Son2.h followed by whatever is in Main.h itself. At this level, the idea of hiding content doesn't make sense. You explicitly asked the preprocessor to put those contents there. It doesn't know anything about the text it's inserting, so it can't follow any kind of Objective-C directive, even if such existed.
The comments are getting to another point. Importing in a .h should be kept to an absolute minimum. The most common reasons to import are for class and protocol type definitions, and Objective-C allows you to forward declare both of these precisely to avoid otherwise necessary imports.
If you have something like:
#interface MyClass : NSObject
#property(nonatomic, strong) MyOtherClass *myOtherClass;
#end
You normally would have to #import "MyOtherClass.h". However, with forward declarations, you can do this instead:
#class MyOtherClass;
And move the #import to the implementation file, which is generally hidden from other files.

Related

Getting "Duplicate Interface Definition" error, definitely has to #import ing header files

I'm helping on an iOS project with lots of methods and definitions common to many different classes in the AppDelegate. So, in each of those classes, in the .h file, I use #import "AppDelegate.h". This works fine until I need access to one of those classes that already imports the AppDelegate into another class that imports AppDelegate. At this point, I get a Duplicate Interface Definition error for AppDelegate.
Ok, so that seems fair. I'm already importing AppDelegate into a file that I'm importing, so AppDelegate is getting imported from two different places. So I remove the AppDelegate line, and everything is fine.
But what happens when I need to import two classes that both need to import AppDelegate?
I have a very specific problem that I'm trying to wrap my head around, and I know it is being caused by something that has to do with this, but I'm not sure what. So I'm hoping if I figure out how I'm supposed to be handling this sort of importing, and sort everything else out, and hope that this solves my problem. So to put this in more concrete terms:
I have ClassA.h, ClassB.h, and ClassC.h. All have #import "AppDelegate.h". When I need to use #import "ClassB.h" in ClassA, I remove the #import "AppDelegate.h" line from ClassA. Everything works smoothly. But what happens if I also need to #import "ClassC.h" into ClassA, and but ClassB and ClassC NEED to have the #import "AppDelegate.h"?
EDIT:
I tried the exact scenario I described above in a clean project, and it built fine, so there is something else at play. But what I can say with certainty is that when this issue came up previously with this project, it was a duplicate interface definition of AppDelegate, and when I removed the #import "AppDelegate.h" line, the error went away, and I still had access to the AppDelegate.h methods and enums through other imported files.
The best prevention and cure for this is to follow some guidelines on when to import from a header file. As a general rule, never import from an Objective-C header except in these cases:
You need to extend a class declared in another header.
You need to declare conformity with a protocol declared in another
header.
You need to refer to a non-class, non-protocol type defined in another header in the public methods and / or properties. To refer to protocols and classes, forward declare them with #class or #protocol, like #class ClassFromOtherHeader;
Every other #import should go in your implementation. My recommendation is to start moving all your #import statements out of headers and into the implementation files according to these rules. Start with the files you think are at the root of the problem and move outward. This will fix your problem and give you the side benefit of clearer code and faster build times.
For me none of the above answers were helping, nor did the answer given here work.
What fixed it for me was closing Xcode, going to ~/Library/Developer/Xcode/DerivedData and deleting all of the derived data associated with this project. After that I reopened the project and it was working fine.
Hope that helps someone!
In my case, none of the solutions mentioned fixed the issue. Xcode was reporting a duplicate interface for a class I rewrote in Swift. Somehow it kept pulling in the Objective-C header file for a class that wasn't directly referenced in the project.
I opened the Terminal, cd into the project directory, then ran the following to track down any files that were including the class header:
grep -nr ProblemClassName.h .
It turned out that the bridging header included an obsolete file that wasn't even referenced in the project navigator. That in turn was importing the header files referenced in the Xcode error, that were also not included in the Xcode project navigator. Now I know to not rely only on the Xcode project navigator for files referenced by the error.
tl;dr Double check the bridging header to ensure that all files that are imported there should be there and are not importing headers that are in-turn importing the problem headers.
I found that a project had a subproject and instead of referencing the includes in the subproject with the proper syntax:
#import <SubProject/Filename.h>
It was directly importing them
#import <Filename.h>
This was only possible because the path of the subproject was included in the "header search paths" of the main project - which is the wrong way to do business. So I deleted it from there.
The subproject should copy the needed included files in its "build phases - copy files" section (which was already happening actually), and the proper form of import that uses the Subproject/Filename.h syntax should be used.
Fwiw I started getting this seemingly at random - for me the fix was to do Product->Clean and it magically went away.
For me, I forgot to include parenthesis in interface definition in m file.

Importing into .h versus .m

Is there a difference between importing something (e.g. #import "JSON.h") into the header file versus the implementation file?
If you #import it in the header, then everything including that header gets it. You may find that useful, in that you don't have to #import it again in other places, but my preference is to #import things only where necessary, to minimize dependencies and make builds faster.
I think if you do it in the header file, you save your self some trouble later on in case you reference a class which is defined in the imported file.
In other words, if you import "JSON.h" in the header file, and there's a JSON class (hypothetically) that you will use in your header file (in the interface), then it would save you from having to do the #class directive at the top. Then your implementation file will also be fine since it would import the header file, which itself imported the "JSON.h" file
Basically I think it would be neater and would be more like objective-c if you import the required files in the interface file (.h). As you've probably noticed, interface files are usually short and concise, allowing you to get a quick glance at what a certain class is about and what it does. If you import your files there, you can also see what files/classes it relies on more easily, saving the implementation file (.m) for the actual 'meat'.

What is the difference between #class and #import

When I compile with the following code there are no errors:
#class RootViewController;
//#import "RootViewController.h"
When I compile with the following code I get an error:
//#class RootViewController;
#import "RootViewController.h"
"error: expected specifier-qualifier-list before 'RootViewController'"
I don't understand what the difference is between the two because I used #import in a similar class and it compiled without errors!
#class is used when you need to know the name of a class in a particular file, but you don't need to know any details about the class (its methods, for example). #import is used when you actually need to use the class (i.e., send it a message).
For example, if you're declaring instance variables in a header file, you can use #class to declare an instance variable of a certain type:
#class MyOtherClass;
#interface MyClass : NSObject
{
MyOtherClass *myIvar;
}
#end
Since you're not using myIvar yet, you don't need to know anything about it except that the type MyOtherClass exists.
However:
#import "MyOtherClass.h"
- (void)doSomething
{
[myIvar doSomethingElse];
}
In this case, you're sending the doSomethingElse message to myIvar; the compiler needs to know that instances of MyOtherClass define this method, so you have to import the header file or the compiler will complain.
Why worry about this?
It mostly has to do with dependencies. When you #import file A into file B, file B becomes dependent upon file A -- that is, if file A changes, you'll have to recompile file B. If you use #class in file B, file B is not dependent on file A, and thus doesn't need to be recompiled when file A changes -- so if you're just declaring a type and not actually dependent upon the implementation of file A, you can save yourself compilation time by not #importing file A.
I decided to refer to the documentation because I was still confused:
#import
This directive is identical to #include, except that it makes sure that the same file is never included more than once. It’s therefore preferred and is used in place of #include in code examples throughout Objective-C–based documentation.
This convention means that every interface file includes, indirectly, the interface files for all inherited classes. When a source module imports a class interface, it gets interfaces for the entire inheritance hierarchy that the class is built upon.
#class
Declarations like this simply use the class name as a type and don’t depend on any details of the class interface (its methods and instance variables), the #class directive gives the compiler sufficient forewarning of what to expect. However, where the interface to a class is actually used (instances created, messages sent), the class interface must be imported.
http://developer.apple.com/iphone/library/documentation/Cocoa/Conceptual/ObjectiveC/Articles/ocDefiningClasses.html#//apple_ref/doc/uid/TP30001163-CH12-TPXREF123
Basic rule: use #class in you header file and #import in your implementation file.
(However, you need to #import your class' superclass. And in some other circumstances you also need to use `#import" in the header.)
#import is not equivalent to #include. If a file is included many times, it will be loaded each time, but with many #imports of the same file, it will still only be loaded once.
Therefore, the main reason to use #class is not to avoid circular dependencies, but to make compilation faster.
Here's an example of when you must use #class
//MYControl.h
#class MYControl; // Must use class
#protocol MYControlDelegate
-(void)control:(MYControl *)control didChangeToState:(UIControlState)state;
#end
#interface MYControl : UIControl
{
id<MYControlDelegate> delegate_;
}
#property (nonatomic, assign) id<MYControlDelegate> delegate;
#end
//MYControl.m
#implementation MYControl
#synthesize delegate = delegate_;
. . .
In this case, there is nothing to import, because the delegate protocol is declared above the main class in the header file. But you still need to be able to refer to the main class which has not yet been declared. So what #class does is to just let the compiler know that there is some class that is called MYControl and will be defined at some point. (Not at runtime however. The class will be defined in the course of the compilation.)
EDIT: From the Objective-C manual:
Since declarations like this simply
use the class name as a type and don’t
depend on any details of the class
interface (its methods and instance
variables), the #class directive gives
the compiler sufficient forewarning of
what to expect. However, where the
interface to a class is actually used
(instances created, messages sent),
the class interface must be imported.
Typically, an interface file uses
#class to declare classes, and the
corresponding implementation file
imports their interfaces (since it
will need to create instances of those
classes or send them messages).
The #class directive minimizes the
amount of code seen by the compiler
and linker, and is therefore the
simplest way to give a forward
declaration of a class name. Being
simple, it avoids potential problems
that may come with importing files
that import still other files. For
example, if one class declares a
statically typed instance variable of
another class, and their two interface
files import each other, neither class
may compile correctly.
Note that circularity is mentioned in the last sentence as one in a general class of issues dealt with by using #class.
#class is used to avoid circular dependency... This prevents circular references where in one header A imports a second header B which(B) imports the first(A) which imports the second (B)and so on in an endless cycle....#class is generally used to ask compiler to look for its definition at runtime... especially when it resides in some static library..
Other than that #import works
See this question
#class:- It defines that you can create instance variable of the imported class and use it in your class.
import:- It defines that you can access the variables declared in the required imported class.
you can use given link for more info.
http://developer.apple.com/iphone/library/documentation/Cocoa/Conceptual/ObjectiveC/Articles/ocDefiningClasses.html#//apple_ref/doc/uid/TP30001163-CH12-TPXREF123
#class means that the definition of the class RootViewController is not yet declared, but will be defined at run time. I believe it is like declaring an extern class in c++.
#import is equivalent to #include.
by the error message i could guess you just made a mistake somewhere inside RootViewController.h, such as a forgotten ; or something like that
you must have imported the this class in the class to which you want to import here. That is why you are getting error , but it can be rectify by #class example .
#class is a forward declaration, a good practice is to put them in the .h instead of #import for avoiding circular #import problem.

Objective-C #import loop

I have the following code:
#import <Foundation/Foundation.h>
#import "ServerRequest.h" // works even though this line is included
#import "ServerResponseRecord.h"
#protocol ServerRequestDelegate<NSObject>
-(void)request:(id)request gotResponseRecord:(ServerResponseRecord*)response;
-(void)request:(id)request gotError:(NSError*)error;
#end
It compiles and runs fine. However, if I replace the method declarations with:
-(void)request:(ServerRequest*)request gotResponseRecord:(ServerResponseRecord*)response;
-(void)request:(ServerRequest*)request gotError:(NSError*)error;
I get the unexpected syntax error "error: expected ')' before 'ServerRequest'". The only reason I can think this might be a problem is that ServerRequestDelegate.h and ServerRequest.h #import each other. However, I don't understand why the code works with the #import line with (id)request. I also don't understand why it's a syntax error.
Can someone provide a good explanation?
You've already hinted at the explanation: an #import cycle.
The first thing I'd do is remove the #include and add the following line above the #protocol definition:
#class ServerRequest;
This is a forward class declaration, and can help break the import loop. Check out this SO question for more details. Apple also has a brief explanation in this guide.
Basically, #import'ing a file causes the compiler to bring the entire text of that file into the file in question, and although #import is "smarter" than #include, it doesn't mean you're immune from import errors. The #class declaration is a way to tell the compiler that a class exists without importing the header. It's appropriate to use when you only need to know about the class name, but don't care about the methods it provides. Generally, you want to use #class in the .h file and #import in the .m file, where you're actually interacting with the class.
#import "loops" are not a problem. #import is the same as #include except that it tracks files and makes sure the preprocessor only reads them in the first time.
Usually when you get an error like that it is due to a problem in the included file. So the error is probably in ServerResponseRecord.h, thought it is probably being tripped by actually using the object declared by it. Without seeing the complete headers it is not possible to say exactly what is going on.

Objective-C: Importing headers in .h or .m?

I'm new to objective-c and would like to know the best practice for importing some external headers that I use in my class.
Should I be storing the #import "classB.h" in my own classes .h file or in the .m file?
What's the difference?
Thanks!
It is proper practice to put a forward class declaration (#class classB;) in the header and #import "classB.h in the .m
A forward class declaration, like #class classB; lets the compiler know it should expect the class later on, and it shouldn't complain about it at the moment.
To avoid circular references, only #import a header file in another class's header file if it's inheriting from that class. Otherwise, use #class ClassName to declare the class type if you need it in your header file, and #import it in the implementation file.
To the compiler, it really doesn't matter. You could just throw forward declarations in your .h and then wait to #import until your .m file. See this post on SO for more info on this.
From a clean-code prospective, some may argue that putting the imports in your implementation file keeps the details closer to where they are needed (see that link above as well; the people there reference this idea).
It's recommended that you import other header files in your header file. That way you can use the class in both the header and the implementation files (because the implementation file (.m) imports its associated header file).
If you want to know when to import files and when to use forward-class declaration, you can go here. ;-)