Here is the basic outline of what I am trying to achieve
class Interface {
public:
virtual ~Interface () {}
virtual void work() = 0;
static Interface *create();
static void setFactory(std::function<Interface *()>);
}
in Interface.cpp I have
static std::function<Interface *()> factory = nullptr;
Interface *Interface ::create() { return factory(); }
void Interface::setFactory(std::function<Interface *()> someFactory) {
factory = someFactory;
}
separately I have
class Implementation : public Interface {
...
}
And I made sure to call
Interface::setFactory([]() -> Interface * { return new Implementation(); });
prior to making any calls to Interface::create(). What I am seeing is that static factory variable is set (no longer a nullptr), so the call to setFactory did work. However resulting factory is empty and calls to Interface::create() would crash. I could probably find a workaround by making a structure with operator() but I wonder if there is an error in my lambda expression.
It is probably worth mentioning that I am using MSVC2013
You may have a static initialization order problem.
My theory is that your factory is assigned to before being constructed, then it is constructed (wiping its state), and then you call create.
MSVC2015 supports magic static local variables, which mostly solves that problem.
namespace {
std::function<Interface *()>& factory() {
static std::function<Interface *()> retval;
return retval;
}
}
now factory() wraps a static local variable that is guaranteed to be constructed at or before the first time you call the function, and only constructed once.
Doing something equivalent in MSVC2013 is harder. If you know that there won't be multiple threads before main starts calling it, you can do the above plus:
static auto&& force_call = factory();
after it to force a call before main starts.
Related
Imagine I have a C++ class called MyClass.
Imagine that I have no access to the source code of MyClass ... it is contained in a library and I am supplied only the library and the header file for MyClass.
Imagine that the class itself requires environment pre-configuration ... for example ... before the constructor of the class can be called, I need to do some setup. The class is normally meant to be used as follows:
void func() {
doGlobalSetup();
MyClass myInstance(1,2,3);
myInstance.doSomething();
...
}
Now I have the situation where we need to create a global instance of the class such as:
MyClass myInstance(1,2,3);
int main(int argc, char *argv[]) {
doGlobalSetup();
myInstance.doSomething();
}
The problem is that in this story, the instance of MyClass is created before the call to doGlobalSetup(). It is instantiated before main() is called. What I want to do is either defer the creation of myInstance() till later or be able to run doGlobalSetup() somehow before the instantiation of the class.
This is a simplification of the actual story ... so let us assume:
I can't change the internals of MyClass.
There must be an instance variable called myInstance of type MyClass (I can't change the logic to MyClass *pMyInstance).
Many thanks for reading.
Since you've constrained the problem such that new cannot be used, you should be able to create the object as always and copy it to the global instance. For example:
MyClass createMyClass()
{
doGlobalSetup();
return MyClass(1, 2, 3);
}
MyClass myInstance = createMyClass();
int main()
{
myInstance.doSomething();
return 0;
}
Does it suit your needs?
namespace
{
int doStaticGlobalSetup()
{
doGlobalSetup();
return 0;
}
}
MyClass myInstance(doStaticGlobalSetup() + 1,2,3);
int main() {
myInstance.doSomething();
return 0;
}
If you absolutely have to defer any constructor calls until after global initialization is done, and want to be sure that no static order initialization fiasco happens, there is a way: make myInstance a reference to uninitialized block of memory and create object in it using placement new after global initializaton.
#include <iostream>
#include <type_traits>
struct foo
{
foo() { std::cout << "created\n"; }
void meow() { std::cout << "used\n"; }
~foo() { std::cout << "destroyed\n"; }
};
void doGlobalSetup() { std::cout << "Global setup\n"; }
//Actual implementation
namespace {
typename std::aligned_storage<sizeof(foo), alignof(foo)>::type bar;
}
foo& instance = reinterpret_cast<foo&>(bar);
//Allows automatic creation and destruction
struct initializer
{
initializer()
{
if (!initialized)
new (&instance) foo();
initialized = true;
}
~initializer()
{
if(initialized)
instance.~foo();
initialized = false;
}
private:
static bool initialized;
};
bool initializer::initialized = false;
int main()
{
doGlobalSetup();
initializer _;
instance.meow();
}
Use a static variable inside a function.
MyClass &myInstance() {
doGlobalSetup();
static MyClass myInstance(1,2,3);
return myInstance;
}
void func() {
myInstance().doSomething();
}
You probably already got the answer you wanted. But just to cover the whole spectrum: if, for some reason, you want to make sure that other places in the code don't accidentally construct MyClass independently of your global variable--and before the global setup has been made--you need to solve this with linking.
If you're on Linux, you can LD_PRELOAD a shared-object containing just the symbol for MyClass's constructor. In it, you declare the setup function accordingly, and let the dynamic linker do the job for you. Then, inside the constructor, you call the setup function, and then do a dlsym("...", RTLD_NEXT) to get the pointer to the original constructor, and you call it, passing it the arguments you got. Of course, you maintain and check a static flag for whether setup has already been performed.
Again, this is probably overkill for you, but I'm posting it just in case someone needs (and is able to use) this kind of solution.
P.S. This is what you get when you rely on global state! :)
First, bear in mind that given a library init function like doGlobalSetup there is a distinct nonzero chance the library will just not work if you create a global instance.
Otherwise it's super easy to create an initializer with the comma operator:
bool do_my_setup = (doGlobalSetup(), true);
MyClass myInstance(1,2,3);
Within the GCC compiler environment there is a function attribute capability called constructor. This allows us to flag a function definition with the ability for it to be automatically invoked before main is called and, most importantly before any class constructors are invoked.
Referring back to the original problem definition ... if the doGlobalSetup() function is modified from:
void doGlobalSetup() { ... }
to
__attribute__((constructor)) void doGlobalSetup() { ... }
then its invocation will occur before main is called and also before any static class instance constructors are called. The explicit call to this function would also be removed from main() as its work has been performed implicitly.
I'm making a game engine and I'm using libraries for various tasks. For example, I use FreeType which needs to be initialized, get the manager and after I don't use it I have to de-initialize it. Of course, it can only be initialized once and can only be de-initialized if it has been initialized.
What I came up with (just an example, not "real" code [but could be valid C++ code]):
class FreeTypeManager
{
private:
FreeTypeManager() {} // Can't be instantiated
static bool initialized;
static TF_Module * module; // I know, I have to declare this in a separate .cpp file and I do
public:
static void Initialize()
{
if (initialized) return;
initialized = true;
FT_Initialize();
FT_CreateModule(module);
}
static void Deinitialize()
{
if (!initialized) return;
initialized = false;
FT_DestroyModule(module);
FT_Deinit();
}
};
And for every manager I create (FreeType, AudioManager, EngineCore, DisplayManager) it's pretty much the same: no instances, just static stuff. I can see this could be a bad design practice to rewrite this skeleton every time. Maybe there's a better solution.
Would it be good to use singletons instead? Or is there a pattern suiting for my problem?
If you still want the singleton approach (which kind of makes sense for manager-type objects), then why not make it a proper singleton, and have a static get function that, if needed, creates the manager object, and have the managers (private) constructor handle the initialization and handle the deinitialization in the destructor (though manager-type objects typically have a lifetime of the whole program, so the destructor will only be called on program exit).
Something like
class FreeTypeManager
{
public:
static FreeTypeManager& get()
{
static FreeTypeManager manager;
return manager;
}
// Other public functions needed by the manager, to load fonts etc.
// Of course non-static
~FreeTypeManager()
{
// Whatever cleanup is needed
}
private:
FreeTypeManager()
{
// Whatever initialization is needed
}
// Whatever private functions and variables are needed
};
If you don't want a singleton, and only have static function in the class, you might as well use a namespace instead. For variables, put them in an anonymous namespace in the implementation (source) file. Or use an opaque structure pointer for the data (a variant of the pimpl idiom).
There's another solution, which isn't exactly singleton pattern, but very related.
class FreeTypeManager
{
public:
FreeTypeManager();
~FreeTypeManager();
};
class SomeOtherClass
{
public:
SomeOtherClass(FreeTypeManager &m) : m(m) {}
private:
FreeTypeManager &m;
};
int main() {
FreeTypeManager m;
...
SomeOtherClass c(m);
}
The solution is to keep it ordinary c++ class, but then just instantiate it at the beginning of main(). This moves initialisation/destruction to a little different place. You'll want to pass references to FreeTypeManager to every class that wants to use it via constructor parameter.
Note that it is important that you use main() instead of some other function; otherwise you get scoping problems which require some thinking how to handle..
I have a framework function which expects an object and a member function pointer (callback), like this:
do_some_work(Object* optr, void (Object::*fptr)()); // will call (optr->*fptr)()
How can I pass a lambda expression to it? Want to do somethink like this:
class MyObject : public Object
{
void mystuff()
{
do_some_work(this, [](){ /* this lambda I want to pass */ });
}
};
The meaning of it all is to not clutter the interface of MyObject class with callbacks.
UPD
I can improve do_some_work in no way because I don't control framework and because actually it isn't one function, there're hundreds of them. Whole framework is based on callbacks of that type. Common usage example without lambdas:
typedef void (Object::*Callback)();
class MyObject : public Object
{
void mystuff()
{
do_some_work(this, (Callback)(MyClass::do_work));
}
void do_work()
{
// here the work is done
}
};
SOLUTION Here's my solution based on Marcelo's answer:
class CallbackWrapper : public Object
{
fptr fptr_;
public:
CallbackWrapper(void (*fptr)()) : fptr_(fptr) { }
void execute()
{
*fptr_();
}
};
class MyObject : public Object
{
void mystuff()
{
CallbackWrapper* do_work = new CallbackWrapper([]()
{
/* this lambda is passed */
});
do_some_work(do_work, (Callback)(CallbackWrapper::execute));
}
};
Since we create the CallbackWrapper we can control it's lifetime for the cases where the callback is used asynchonously. Thanks to all.
This is impossible. The construct (optr->*fptr)() requires that fptr be a pointer-to-member. If do_some_work is under your control, change it to take something that's compatible with a lambda function, such as std::function<void()> or a parameterised type. If it's a legacy framework that isn't under your control, you may be able to wrap it, if it's a function template, e.g.:
template <typename Object>
do_some_work(Object* optr, void (Object::*fptr)());
Then, you can implement a wrapper template:
template <typename F>
void do_some_work(F f) {
struct S {
F f;
S(F f) : f(f) { }
void call() { f(); delete this; }
};
S* lamf = new S(f);
do_some_work(lamf, &S::call);
}
class MyObject // You probably don't need this class anymore.
{
void mystuff()
{
do_some_work([](){ /* Do your thing... */ });
}
};
Edit: If do_some_work completes asynchronously, you must allocate lamf on the heap. I've amended the above code accordingly, just to be on the safe side. Thanks to #David Rodriguez for pointing this out.
There are deeper problems with the approach that you are trying to take than the syntactical mismatch. As DeadMG suggests, the best solution is to improve the interface of do_some_work to take a functor of some sort (std::function<void()> in C++11 or with boost, or even a generic F on which operator() is called.
The solution provided by Marcelo solves the syntactical mismatch, but because the library takes the first element by pointer, it is the responsibility of the caller to ensure that the object will be alive when the callback is executed. Assuming that the callback is asynchronous, the problem with his solution (and other similar alternatives) is that the object can potentially be destroyed before the callback is executed, causing undefined behavior.
I would suggest that you use some form of plimp idiom, where the goal in this case would be to hide the need for callbacks (because the rest of the implementation might not need to be hidden you could use just another class to handle the callbacks but store it by value, if you don't want do have to dynamically allocate more memory):
class MyClass;
class MyClassCallbacks {
MyClass* ptr;
public:
MyClassCallbacks( MyClass* ptr ) : ptr(ptr) {}
// callbacks that execute code on `ptr`
void callback1() {
// do some operations
// update *ptr
}
};
class MyClass {
MyClassCallbacks callbackHandler;
public:
void mystuff() {
do_some_work( &callbackHandler, &MyClassHandler::callback1 );
}
};
In this design, the two classes are separated but represent a unique single entity, so it is fine to add a friend declaration and let MyClassCallbacks access the internal data in MyClass (both of them are one single entity, divided only to provide a cleaner interface, but coupling is already high, so adding the extra coupling requiered by friend is no problem).
Because there is a 1-1 relationship between MyClass and MyClassCallbacks instances, their lifetimes are bound and there would be no lifetime issues, except during destruction. During destruction you must ensure that there is no callback registered that can kick in while the MyClass object is being destroyed.
Since you are at it, you might want to walk the extra mile and do a proper pimpl: move all of the data and implementation into a different type that is held by pointer, and offer a MyClass that stores a pointer and offers just the public functions, implemented as forwarders to the pimpl object. This could be somehow tricky as you are using inheritance, and the pimpl idiom is a bit cumbersome to implement on type hierarchies (if you need to extend MyClass, deriving from Object could be done in the pimpl object, rather than the interface type).
I don't think you can do that. Your do_some_work() is declared to accept pointer to methods of class Object, so such should be provided. Otherwise optr->*fptr is invalid since the lambda is not member of Object. Probably you should try using std::function and adding the needed members of Object in its closure.
You must use std::function<void()>. Both function and member function pointers are highly unsuited to being callbacks.
I am using a singleton class with a thread that calls into the singleton. I was asked during a review why I used the this pointer instead of the singleton instance.
My code with the suggested changes.
class myClass : public threadWrapper
{
public:
static myClass& instance()
{
static myClass instance;
return instance;
}
// This is the callback that I have implemented
static void callback(void *me)
{
if (me != NULL)
static_cast<myClass*>(me)->doCallback();
}
// This is the suggested callback
static void callback2(void *me)
{
instance().doCallback();
}
// caller gets instance and then calls initialise()
int initialise()
{
if (initialised)
return ERROR_ALREADY_INITIALISED;
// Initialise the class
// my thread startup call
// thread wrapper class initialisation that calls pthread_create that runs the callback method with this as a parameter
// priority is a global value that difines the relative priority of the various threads.
threadWrapper::Initialise(priority, callback, this);
initialised = true;
}
private:
myClass() : initialised(false) {;}
void doCallback(void);
bool initialised;
static const int
}
So is there any significant difference in speed between the two?
The threadWrapper is mandated in the existing code base, and I'm not allowed to use boost.
My justification was that if we needed to make this not a singleton then fewer changes would be required.
The speed difference will be pretty much nonexistent.
As for code quality, Singletons are quite horrendous and I personally would chuck out both forms, especially in a threaded environment. Assuming that it's too late for that, however.
The thing is, if you're gonna pass in a pointer to the object, why not just not make that object global in the first place? And if you are, it should at least be strongly typed. And then, you're just ... wrapping a member method in a static method? Why bother? Anyone who has a pointer to the class can just call the method on it in the first place. This is just insane.
Edit: If you're stuck with the existing design, then the second version is definitely better than the first and no slower. Even if you have existing code that depends on the Singleton, then it's absolutely better to refactor what you can to not depend on it.
If I have two static variables in different compilation units, then their initialization order is not defined. This lesson is well learned.
The question I have: are all the static variables already allocated, when the first one is being initialized. In other words:
static A global_a; // in compilation unit 1
static B global_b; // in compilation unit 2
struct A {
A() { b_ptr = &global_b; }
B *b_ptr;
void f() { b_ptr->do_something(); }
}
int main() {
global_a.f();
}
Will b_ptr point to a valid piece of memory, where B is allocated and initialized at the time of the execution of the main function? On all the platforms?
Longer story:
The compilation unit 1 is Qt library.
The other one is my application. I have couple QObject derived classes, that I need to be able to instantiate by the class name string. For this I came up with a templated factory class:
class AbstractFactory {
public:
virtual QObject *create() = 0;
static QMap<const QMetaObject *, AbstractFactory *> m_Map;
}
QMap<const QMetaObject *, AbstractFactory *> AbstractFactory::m_Map; //in .cpp
template <class T>
class ConcreteFactory: public AbstractFactory {
public:
ConcreteFactory() { AbstractFactory::m_Map[&T::staticMetaObject] = this; }
QObject *create() { return new T(); }
}
#define FACTORY(TYPE) static ConcreteFactory < TYPE > m_Factory;
Then I add this macro on every QObject subclass definition:
class Subclass : public QObject {
Q_OBJECT;
FACTORY(Subclass);
}
Finally I can instantiate a class by the type name:
QObject *create(const QString &type) {
foreach (const QMetaObect *meta, AbstractFactory::m_Map.keys() {
if (meta->className() == type) {
return AbstractFactory::m_Map[meta]->create();
}
}
return 0;
}
So the class gets a static QMetaObject instance: Subclass::staticMetaObject from the Qt library - it is auto-generated in Q_OBJECT macro I think. And then the FACTORY macro creates a static ConcreteFactory< Subclass > instance. ConcreteFactory in its constructor tries to reference of Subclass::staticMetaObject.
And I was pretty happy with this implementation on linux (gcc), until I compiled it with Visual Studio 2008. For some reason AbstractFactory::m_Map was empty on the runtime, and the debugger would not break at the factory constructor.
So this is where the smell of static vars referencing other static vars is coming from.
How can I optimize this code to avoid all these traps?
Yes, the standard allows this.
There are a number of paragraphs in section [basic.life] which start out
Before the lifetime of an object has
started but after the storage which
the object will occupy has been
allocated or, after the lifetime of an
object has ended and before the
storage which the object occupied is
reused or released, any pointer that
refers to the storage location where
the object will be or was located may
be used but only in limited ways.
and there is a footnote which indicates that this specifically applies to your situation
For example, before the construction of a global object of non-POD class type
Short Answer: Its should work as you have coded it. See Ben Voigt Answer
Long Answer:
Do something like this:
Rather than let the compiler decide when globals are created, create them via static methods (with static function variables). This means that they will be deterministically created on first use (and destroyed in the reverse order of creation).
Even if one global uses another during its construction using this method gurantees that they will be created in the order required and thus be available for usage by the other (watch out for loops).
struct A
{
// Rather than an explicit global use
// a static method thus creation of the value is on first use
// and not at all if you don't want it.
static A& getGlobalA()
{
static A instance; // created on first use (destroyed on application exit)
// ^^^^^^ Note the use of static here.
return instance; // return a reference.
}
private:
A()
:b_ref(B::getGlobalB()) // Do the same for B
{} // If B has not been created it will be
// created by this call, thus guaranteeing
// it is available for use by this object
}
B& b_ref;
public:
void f() { b_ref.do_something(); }
};
int main() {
a::getGlobalA().f();
}
Though a word of warning.
Globals are an indication of bad design.
Globals that depend on other globals is another code smell (especially during construction/destruction).
Yes. All are located in .data section, that is allocated at once (and it's not heap).
Putting it another way: if you are able to take its address, then it's OK, because it surely won't change.
If B has a constructor, like A has, then the order that they are called is undefined. So your code won't work unless you are lucky. But if B doesn't require any code to initialise it, then your code will work. It's not implementation-defined.
Ah, but the idea that static variables are "not initialised" is quite wrong. They're always initialised, just not necessarily with your initialiser. In particular, all static variables are created with value zero before any other initialisation. For class objects, the members are zero. Therefore global_a.b_ptr above will always be a valid pointer, initially NULL and later &global_b. The effect of this is that use of non-pointers is unspecified, not undefined, in particular this code is well defined (in C):
// unit 1
int a = b + 1;
// unit 2
int b = a + 1;
main ... printf("%d\n", a + b); // 3 for sure
The zero initialisation guarantee is used with this pattern:
int get_x() {
static int init;
static int x;
if(init) return x;
else { x = some_calc(); init = 1; return x; }
}
which assures either a non-return due to infinite recursion, or, correctly initialised value.