Entries in iPhone (3)

Friday
Aug132010

OCPtr - a Smart Pointer for Objective C

In my last post I covered why we might want Garbage Collection on the iPhone, some reasons why we don't have it, and how the same problem is solved in C++. I then hinted that we might be able to bring the same C++ goodness to Objective-C - if we allow ourselves to use Objective-C++.

In this post I'm going to introduce my own solution.

Say hello to OCPtr

Rather than dive into the implementation, let's look at usage - and how it addresses our memory management needs. Let's start with our first example - allocating an NSString. Here's how it looks using OCPtr:

OCPtr<NSString> str = [[NSString alloc] initWithFormat: @"One: %d", 1];

// ...

The first thing to notice here is that instead of declaring our type as NSString* we declare it as OCPtr<NSString>. This is how smart pointers work. C++'s power is with types - especially when mixed with templates. OCPtr<NSString> is a complete type. Furthermore it is a value type. At the end of the scope it will be destroyed. In its destructor we call release on its NSString* member - which brings us to the second thing to notice - we have omitted the [str release] step!

Let's think about that for a moment. On the face of it we have saved ourselves 14 characters - but at the cost of six characters in the declaration - a net saving of eight characters. Woohoo! Obviously if this was all there was to it I wouldn't be writing this blog post. I'm not that obsessed with removing redundancy!

What is more significant is the reduction in the mental overhead of tracking when you need to call release - and the debugging overhead when it goes wrong - not to mention the potential for financial overhead if it only goes wrong when your users are using it.

In this simple example it doesn't seem to have given us much - but think about this again next time you're trying to track down a particularly elusive leak or over-release.

Of course real code is more complex than this, and OCPtr will need to be more than this to function as a stand-in for raw pointers. Years of C++ smart pointer experience tells us we need to overload the assignment operator, provide copy constructors (allowing us to create one OCPtr from another), and ideally overload a few other operators too. Technically a true smart pointer would overload the -> and & operators to behave like raw pointers - but these are not used with Objective-C types, so I haven't provided them. Other than that OCPtr provides all of this, so simple assignments result in retain counts being updated appropriately, and objects are released as their OCPtr wrappers leave the scope.

Some smart pointers also overload the conversion operator (this is invoked when you try to cast an object - whether explicitly or implicitly) to convert to the underlying raw point. This is a controversial practice, and can lead to ambiguities. However in the constrained environment of Objective-C code it seems safe enough - and gives us one key advantage: it allows us to send messages to the underlying object without additional syntax. Here's some more code illustrating this as well as some of the other preceding points:

{
  OCPtr<NSString> str = [[NSString alloc] initWithFormat: @"One: %d", 1];

  OCPtr<NSString> str2;
  {
    OCPtr<NSString> str3 = str; // retain count == 2
    str2 = str3; // retain count == 3

  } // retain count == 2

  // ...

} // retain count == 0 - dealloc called

Not just Smart - it smells nice too

So what else can we do with OCPtr?

Well it wouldn't be so useful if it couldn't also manage instance variables. OCPtr does that too - as long as you have the project setting enabled for: "Call C++ Default Ctors/Dtors in Objective-C". This setting seems to be enabled by default now, but it's worth checking. Now if all your ivars are held as OCPtrs you don't even need to write a dealloc method. The compiler will effectively write it for you - and all your OCPtr destructors will be called - releasing all their managed objects.

What about properties?

To some extent there is an overlap between the functionality of synthesised properties and what we are doing here. That is, if you declare a property for an Objective-C object with the 'retain' attribute then the synthesised code will include the retain-release code necessary to do the right thing. This does relieve the programmer of some work - so is OCPtr buying us anything?

Well, using properties still places two responsibilities on the caller: First they must remember to use the property syntax (whether the dot-syntax or the direct messaging passing) everywhere (including within the implementing class - except at the point of allocation). Secondly they must still remember to set the properties to nil in dealloc - which is really no improvement over just releasing them.

So retain properties do help with change of ownership - but at the cost of having two ways to do so - one which is automatic - the other still manual.

But can we declare properties for OCPtr ivars - and what happens with the retain counts?

Well recall that you can write a property for any Objective-C type - which includes all C types - primitives and structs. In the latter case you will use the assign attribute instead of retain (and assign is the default - so even easier). In this case no retain-release code will be generated - values will just be assigned directly. But OCPtr overloads assignment to provide retained change of ownership semantics. This gives us exactly what we want!

@property OCPtr<NSString> str;

So, in short, use assign properties to provide external access to OCPtr ivars. Internally you can use either form of access but they work the same way. In either case assignment works as it should and references will be cleaned up in dealloc. This is surely an improvement.

Transfer of ownership

How does OCPtr work with autoreleased objects? If you initialise an OCPtr with an autoreleased object you do still have to tell it to take ownership. This is no different to a raw Objective-C pointer. In both cases to take ownership you must call retain. E.g:

OCPtr<NSString> str = [[NSString stringWithFormat: @"One: %d", 1] retain];

It niggles me a bit that we still have to make a distinction between autoreleased and non-autoreleased objects - but that's a property of the autorelease mechanism itself, rather than a limitation of OCPtr (I could imagine a scheme where OCPtr detected that the pointer being passed was in the autorelease pool - but there would be no way to know for sure if that related to this assignment - and would probably be taking things to far anyway).

But take a step back for a moment. What is the problem that autorelease solves in the first place? autorelease's raison d'ĂȘtre is to provide transfer of ownership semantics. If a method creates an object but does not own it (typically a factory method) - it just returns it. In order to return it with a valid retain count, without requiring that the caller release it, it adds it to the autorelease pool to be released some time later. This works, but adds extra rules and syntax, and can result in objects living longer than they need to.

But smart pointers already solve this problem. By returning an OCPtr, the retain count will remain valid until after the assignment operator of the caller's OCPtr has retained it. The returned OCPtr then immediately goes out of scope (it is a temporary object). The net result is that you can freely create and return OCPtrs just as you can primitive types - no need to worry about retain counts or autorelease pools.

-(OCPtr<NSString>) someMethod
{
    OCPtr str = [[NSString alloc] initWithFormat: @"created here"];
    return str;

} // str will be released here but caller already has it

// ...

-(void) someOtherMethod
{
    OCPtr<NSString> str = [self someMethod];

} // str will be released here, retain will go to 0 and dealloc called

Of course if you're working with third-party APIs (including the SDK frameworks) you will still need to work with autoreleased objects at times, so it's worth remembering that you still need to retain them before putting them in an OCPtr.

But wait - there's more

What we have discussed so far covers our memory management needs. But if we've accepted a general purpose object wrapper into our code we have opportunity for further, aspect-oriented, benefits:

id type checking

One of the great things about Objective-C is that it is a dynamic language (albeit with static underpinnings).

One of the biggest problems with Objective-C is that it is a dynamic language (fortunately with static underpinnings).

While it is nice that we can choose to use typed, untyped (id) or partially typed (NSObject) objects, sometimes we are forced to go untyped when we'd like the benefit of type checking. The consequences of getting the types wrong are usually crashes, with not so helpful error messages - and at a different point in the code. We can check types at runtime, of course, with the isKindOfClass: method, to which you have to pass a class object - obtained by passing the class message. This can clutter the code up with mechanics.

OCPtr provides a conversion constructor from both id and NSObject, which will test the types (using the class and isKindOfClass: methods) before casting internally. As a result if we do this:

OCPtr<NSArray> str = [[NSString alloc] initWithFormat: @"hello"];

... we will get an exception that tells us exactly what happened.

If we had used a raw NSArray* pointer here the assignment would have worked - but we'd get errors further down the line if we tried to call array methods on it. These can be difficult to track down.

Don't want to pay for the check? Just cast to the target type before you assign (but lose the benefit of the type checking - so the principle is "correct by default, fast where necessary").

Release early

By eliminating your memory management bugs you will be able to release your apps earlier - but actually I was referring to releasing objects early.

Sometimes you're done with an object in the middle of a scope and you want to release it there and then. If you do this you are strongly advised to then set it to nil - to avoid the chance of anyone trying to use no longer valid memory. With an OCPtr you need only set it to nil and you get both advantages. You've been able to do this for a while with properties, but now you can do it directly with ivars, and even with local variables:

{
	OCPtr<NSString> str = [[NSString alloc] initWithFormat: @"One: %d", 1];

	// ...

	str = nil; // release is called here, and the underlying pointer set to nil

	// ...
}

Logging hooks

A powerful, but easily abused, feature of C++ templates is a concept known as specialisation (unfortunately a rather overloaded term in OO languages). A template is specialised when you write specific code for the case where a template's type argument(s) are of specific types. If that doesn't make things any clearer I'll explain how this relates to OCPtr and logging and hopefully it will click.

OCPtr comes with another template class: OCPtrHooks. OCPtrHooks declares a set of empty methods and nothing else. Each method represents a state change in OCPtr (e.g. onAssign) and OCPtr uses an OCPtrHooks class, parameterised with the same Objective-C type, calling the appropriate hook method as things happen. Because all the methods are empty the compiler is able to optimise these calls away completely.

So if the methods do nothing and they are not even compiled in what use is this? Well, due to the magic of template specialisation we can write a version of OCPtrHooks specialised for a particular type - or even partially specialised for a base type. Then, for those specialised types only, your custom versions will be called.

You can implement your specialisations to do anything - but a useful implementation for us is to log the events. Enabling logging for a particular type is as easy as declaring a specialisation that derives from a logging base class, like this:

template<> class OCPtrHooks<NSString> : public OCPtrLogHooks{};

Don't worry about trying to follow that if you're not a C++ programmer. The important bits are the template type parameter (NSString, here) and the base class (OCPtrLogHooks). Just substitute the NSString for any type you want to log and it will start working - with no overhead (not even an if) in all other cases.

While this is powerful, and useful, it does make use of, and expose, some tricky template syntax - If you're not already familiar enough with C++ to know how this works you may choose not to take advantage of this facility (I might try and make it friendlier in the future - even if that involves the use of a wrapper macro).

The dark side

So we've eliminated the mental overhead of manual retain counts, without introducing any runtime overhead, added transparent runtime type checking to dynamic types, along with several other benefits. With all this goodness everyone will want to use OCPtr, right? They'd be mad not to?

Well, that's generally true of C++ smart pointers in the C++ world. But because we're intruding the world of C++ into the world of Objective-C, and using a hybrid language to do so, there are some drawbacks to consider. These are the ones I think are relevant:

  1. OCPtr is a C++ template. This means it must be #included or #imported - so all your implementation files will need to be .mm files (or you can set the filetype in the file info dialog).
  2. C++ syntax intrudes into your application code in the form of the OCPtr<MyClass> syntax.
  3. The idea of custom value types may be unfamiliar to people reading the code. The fact that assignments handle retain counts for you may be a surprise, for example.
  4. If you use an OCPtr in an untyped context, e.g. as an argument to NSLog, the compiler cannot deduce that it needs to return the raw pointer out. So you'll need to explicitly access it - either calling a C++ method, such as get(), or casting to the underlying type.

Issues 1 & 2 are the most likely to put someone off - and they become especially significant if you are writing code for a client, or as part of a team in a larger company - especially if you are not already using any Objective-C++ on the project.

So I wouldn't necessarily recommend that everyone just start using OCPtr everywhere - but if you are just writing for yourself - or as a small, open-minded, team - I'd encourage you to at least give it a try and see if it can make your life easier.

But at the end of the day, even if you decide the trade-offs are not worth it for you, you can at least rest easy knowing that manual referencing counting is a choice. And you can tell all your gloating friends who use other languages, "I don't need no stinking garbage collection!"

Give me the codez

So where can you get OCPtr. I'll shortly be putting it up on GitHub. When I do so I'll update here with a link.

Friday
Aug132010

We Don't Need No Stinking Garbage Collection

iPhone developers are sometimes called names or excluded from polite circles for not having Garbage Collection. Recently I gave a presentation at the London iPhone Developers Group meeting which covered this matter from a few angles - including some techniques of my own. There was enough interest that I thought I should write it up in a more widely distributed format - where I could also expand on some points I glossed over due to time constraints (it was 20 minute slot).

Koviks avfallsdeponi.jpg

What's that Smell?

So what's the stink about? Well, if you develop for the iPhone you're going to have to use Objective-C (at least for some of your code - e.g. the UI). Until recently Objective-C has not had any form of Garbage Collection. Memory must be managed manually (although helped by a reference counting system). As of Objective-C 2.0 there is now a GC facility but this is only available on Mac OSX - from Snow Leopard on.

Most languages in common use today have Garbage Collection. There are three stand-out exceptions: Objective-C (on the iPhone), C and C++. As it happens this is precisely the same trio that are sanctified for use on the iPhone (I'm deliberately avoiding the issue of front-end languages/ platforms such as Monotouch - especially while their status with respect to the developer licence remains in doubt)!

So why no GC on the iPhone? Before we look at that I think it's worth a quick review of why anyone would think it was needed in the first place.

Remembering to forget your memory usage

Consider this typical snippet of Objective-C code:

NSString* str = [[NSString alloc] initWithFormat: @"One: %d", 1];

// ...

[str release];

The use of NSString here is not that interesting. What we're looking at is that, because we used alloc to get the string we need to remember to call release when we're done. Note that release doesn't necessarily deallocate the memory for the string there and then. The Objective-C runtime uses a reference counting (or retain counting, in Obj-C parlance) system to track when the last reference is released - at which point dealloc will be called - giving the object a chance to clean up its own resources.

This is hardly rocket science, and already has many advantages over the raw C way of doing things where ownership can be passed around like a hot potato and often it is difficult to know if you should be responsible for freeing or not (yes - conventions exist to mitigate this - but they are not standardised).

But when you start putting stretches of code in the middle, maybe with multiple exit points (unless you're a SESE fanatic - and eschew exceptions) it already starts to add mental overhead. Not much, maybe, especially to an experienced developer - but spread across thousands of instances it adds up. When you're writing code you want to focus as much as possible close to the domain level of abstraction - and these language mechanics issues can detract from that.

As well as the pattern shown in the example above there are other variations to consider. If the object is an instance variable you have to put the release in the dealloc method (or set it to nil via a property if you prefer). If the object is obtained from a static convenience constructor (such as [NSString stringWithFormat:]), or if you send it the autorelease message, then you should not call release yourself - but you do need to be aware the the object will live beyond its last use - which may be significant.

If you are given an object, but not passed ownership, and need to keep hold of it then you will need to send it the retain message (then later remember to call release again).

Whichever case it may be, doing it wrong has consequences. Failing to call release will result in leaks (which may lead to a crash - but much later). Failing to call retain could mean that you are using an object that may have been deallocated. This will likely lead to a crash sooner. If you're really lucky it will crash close enough to the source of the problem to help you find it - unless your users find it first.

Leaks and crashes are serious problems. Even if 99% of your code is correct in its memory management, just one or two such bugs can ruin your user experience (not to mention your app store ratings).

Deodorant

There are some tools that can help. For leaks we now have the LLVM static analyser integrated with XCode. This will work with the compiler to analyse your code paths, looking for misplaced releases or retains. It does a pretty good job and I highly recommend using it regularly. But its not perfect. It misses quite a few cases - especially if the code paths get complex. It can also give false negatives.

At runtime we can also use the Leaks instrument, which will track your references and see when objects are still alive after their last use. I suspect that internally it implements garbage collection, or something like it, to do the reference tracking. Instruments itself has got very good lately - making it much easier to filter out all the noise and see just your bits of the code. Again I highly recommend using this tool. But again it won't catch everything. In particular it will only test the code paths you use while its running.

For over releases, or releasing too early, we can check stack dumps. This may help us to track down the source of a crash - if it's not too far from the suspect code. A better tool, though, is NSZombies - enabled with the NSZombiesEnabled environment variable. With this in operation objects that would normally be dealloc'ed get transmuted into zombie objects instead. These will detect if you send any messages to them and log the fact. There are also a handful of other tools and techniques for tracking down leaks and over releases after the event.

So we can cover up the smell to some extent - but just as with real smells, masking is not the same as removing. We have additional mental overhead distracting you from your task - and extra tools and techniques required to apply after the event.

That's rubbish

So why is Garbage Collection not provided for iOS, given that it's been available on the Mac for about three years?

The short answer is: performance. The slightly longer answer is that GC has an overhead. It's an overhead that you pay every time you run your app. Much of the time the overhead may be invisible, or at least barely noticeable. Sometimes, though, it becomes very noticeable. Common tasks such as scrolling through long lists are famously smooth user experiences on iOS devices. Not so on other platforms, which can tend to be glitchy and jerky.

While bad code is something that can plague any platform (and there are other factors, such as hardware acceleration)- at least it is something you control. Issues caused by GC, though, are outside your immediate control. Depending on the GC implementation you may have access to an API that allows you some degree of control - but this is usually in the form of hints.

Whether you are willing to pay the cost or not, whether you think it's an issue or not, others clearly do have issues with it (and they may be your users). So Apple, as platform provider, has taken the design decision to hold back on providing garbage collection.

This leaves us with managing memory in our own apps. Notice I didn't say "manually managing memory" there. Any good developer - faced with implementing the same patterns of code over and over - even small snippets of code - but especially where missing them is easy and dangerous - find ways to factor out the duplication.

This is a problem that was solved years ago in C++, employing a concept known as RAII.

RAII of light

What is RAII, and how can it help us in Objective-C?

RAII stands for Resource Acquisition Is Initialisation - which is a long and complicated sounding name for a simple but powerful concept. It's also an unfortunate name as it sounds like the focus is on "acquisition" and "initialisation", yet the most interesting part is at the end of the lifetime. To explain what I mean let me cast it in Objective-C terms (if you're familiar with C++ and this concept feel free to skip ahead):

An Objective-C class has one or more initialisers, and a dealloc method. We use these methods to manage our resources (usually pointers to other objects).

In our init method (or designated initialiser) we typically either allocate other objects, or retain objects passed in. Either way we own references to them that we must release later, in our dealloc method.

Of course, we never call dealloc directly - it is called by release once the last retain count has gone.

In C++ the equivalent of init methods are called constructors and the analog to dealloc is the destructor. Where C++ differs is that instead of all objects being reference counted heap objects we have either plain heap objects or value types (which typically live on the stack or, if they are member (instance) variables they are scoped by the lifetime of the object they are a member of).

The nice thing about value types is that they are destroyed immediately, and predictably, at the end of their scope. This feature is so important it has its own name: Deterministic Destruction.

{
	MyClass myObject;

	// ...

} // destructor will be called here

Now because the the destructor is called automatically at the end of the scope - regardless of how it leaves that scope (so it allows for early returns or exceptions), and we can write arbitrary code in the destructor, this gives us a hook. We can use this mechanism to write code that will be be called as the execution leaves a scope. In fact this technique, along with templates and a bit of operator overloading, has been used for years to implement wrapper classes for raw C++ pointers that look and feel like pointers but have some sort of automatic memory management mixed in. These classes are known, collectively, as smart pointers. Reference counting smart pointers are just one possibility but there are many smart pointer traits that can be captured this way - such is the flexibility (and complexity) of C++.

This is all well and good - but we don't have custom value types, templates or operator overloading in Objective-C. So what use has this discussion been?

Objective C++

Apple's officially sanctioned languages for iPhone development are Objective-C, C and C++. But actually there is a fourth language, not listed there because it's a hybrid. That "language" is Objective-C++. I use the word language lightly here because Objective-C++ is a more of a glue language than a language you would use in its own right. Typically it is used to bridge the worlds between pure Objective-C and pure C++ in order to keep them at arms length (pun intended) from each other.

But there is no technical reason that you couldn't write an entire application, top-to-bottom, in Objective-C++. The reason you wouldn't typically do so is that they have very different syntaxes, strengths and weaknesses, and design trade-offs. Mixing the two just doesn't look like code written in a single language. C++ is oil to Objective-C's water (and we'll resist the temptation to point out how big the leaks can get if you mix oil and water!).

We're going carve out a new niche here. We're not going to freely mix the languages without constraint. But we're not going to keep them siloed either. Instead we're going to use some judicious helpers from the C++ side to assist our otherwise pure Objective-C. I don't suggest this lightly. There are certainly some that would take issue with this approach. We'll discuss these trade-offs later.

In the follow-on article I'm going to put this all together and show you my Objective-C++ solution: OCPtr.

Friday
Apr092010

iPhone Multitasking Revisited

Back in January I posted to this blog my thoughts on what form background processing might take in iPhone OS. I proposed that we would not see what you might consider, "full multitasking", but went on to explain why I didn't think we needed it (and that that would be a good thing) and what should happen instead. Yesterday Apple announced iPhone OS 4 and the headline feature, of course, was multitasking. So how did my musings (and those of the commenters to that article) stack up?

slide_5923_79686_large.jpg

I listed three primary areas that most multitasking usages involve:

1. Background apps, such as music players. The iPod app does run in the background, of course, but there is certainly a case for third-party apps such as Spotify or Last.fm to have the same ability.

This was the first multi-tasking service discussed - the background audio service. Pandora was the poster child.

2. Notifications. This is at least partly addressed by the Push Notification service that Apple now provides. More on that in a moment.

I went on to discuss in more depth what I meant by this and Apple have implemented this exactly as I proposed (although it is conceivable that they arrived at this concept independently). They have called it Local Notifications.

3. Fast switching between apps

I also went on to suggest that this is what most people want most of the time they lament the lack of multi-tasking. App switching has always been fairly smooth on the iPhone and iPad, but new services to allow apps to persist and restore their state quickly make that process really seamless - and even allow apps to respond to beings switched back to (the example given was of playing Tap Tap Revenge which, when switched back to after switching away, has paused the game and gives a 3 second countdown before resuming again).

I felt that these were probably the biggest uses of multi-tasking that could be provided without disrupting the performance and battery life. They are not the only areas, of course. I invited people to suggest other areas that could be served in a similar way.

The first comment I received was from Graeme Foster, who said:

Location-tracking apps like Google Latitude and FourSquare?

And the second, from James Webster, said:

Running the GPS and a background app would be battery intensive

I'm sure Apple were reading these comments. They have addressed both of these concerns. There are two background location services. One is based on cell tower triangulation and is aimed at Graeme's use case. The other is full on location services - aimed at navigation apps - which assume the device will be connected to power - so acknowledging James' concern.

One use, which wasn't mentioned by me or my commenters, but I did see discussed elsewhere, was for VoIP apps, such as Skype - and it was nice to see this catered for too. In fact I wonder if it becomes possible to use an iPod touch just like a normal phone now!

Another feature that I didn't bring out (but others did) was completion of upload tasks. In retrospect this is perhaps the most important from my own perspective. In another iPad related blog post I alluded to an app that I'm working on that fits the iPad like a glove (no, not literally). I'm still holding my cards quite close on that one, but what I will say is that it involves cloud syncing. The trouble with cloud sync enabled apps right now on iPhone OS is that the syncing tends to happen in the background. If the user closes the app while the syncing is in progress then, at best, the app will not be fully synced up and the "cloud" version will be out of date. At worst, if not carefully designed, data corruption may result.

iPhone OS 4 includes a service called Task Completion and this solves the problem cleanly. Tasks such as syncing and uploads that are kicked off while the app is running can continue to completion in the background even after the app is closed. This was one area that I was agonising over for my app and I'm really pleased that it is solved in OS 4.

Finally, one of my commenters, Yitz, posted an interesting link to his own blog with his views on another "iPhoneOS multi-tasking alternative". In fact the bulk of his post was arguably not about multi-tasking as such, but rather about allowing developers to write services that other apps could call into. In retrospect this is now somewhat provided for by the QuickLook framework.

So, in summary, Apple's multitasking features include everything I expected, everything I hoped for and everything I dared not dream of (or, indeed, blog about).

Now things get interesting.