Low level details of inheritance and polymorphism - c++

This question is one of the big doubts that looms around my head and is also hard to describe it in terms of words . Some times it seems obvious and sometimes a tough one to crack.So the question goes like this::
class Base{
public:
int a_number;
Base(){}
virtual void function1() {}
virtual void function2() {}
void function3() {}
};
class Derived:public Base{
public:
Derived():Base() {}
void function1() {cout &lt&lt "Derived from Base" &lt&lt endl;
virtual void function4() {cout &lt&lt "Only in derived" &lt&lt endl;}
};
int main(){
Derived *der_ptr = new Derived();
Base *b_ptr = der_ptr; // As just address is being passed , b_ptr points to derived object
b_ptr -> function4(); // Will Give Compilation ERROR!!
b_ptr -> function1(); // Calls the Derived class overridden method
return 0;
}
Q1. Though b_ptr is pointing to Derived object, to which VTABLE it accesses and HOW ? as b_ptr -> function4() gives compilation error. Or is it that b_ptr can access only upto that size of Base class VTABLE in Derived VTABLE?
Q2. Since the memory layout of the Derived must be (Base,Derived) , is the VTABLE of the Base class also included in the memory layout of the Derived class?
Q3. Since the function1 and function2 of base class Vtable points to the Base class implementation and function2 of Derived class points to function2 of Base class, Is there really a need of VTABLE in the Base class?? (This might be the dumbest question I can ever ask, but still I am in doubt about this in my present state and the answer must be related to answer of Q1 :) )
Please Comment.
Thanks for the patience.

As a further illustration, here is a C version of your C++ program, showing vtables and all.
#include <stdlib.h>
#include <stdio.h>
typedef struct Base Base;
struct Base_vtable_layout{
void (*function1)(Base*);
void (*function2)(Base*);
};
struct Base{
struct Base_vtable_layout* vtable_ptr;
int a_number;
};
void Base_function1(Base* this){}
void Base_function2(Base* this){}
void Base_function3(Base* this){}
struct Base_vtable_layout Base_vtable = {
&Base_function1,
&Base_function2
};
void Base_Base(Base* this){
this->vtable_ptr = &Base_vtable;
};
Base* new_Base(){
Base *res = (Base*)malloc(sizeof(Base));
Base_Base(res);
return res;
}
typedef struct Derived Derived;
struct Derived_vtable_layout{
struct Base_vtable_layout base;
void (*function4)(Derived*);
};
struct Derived{
struct Base base;
};
void Derived_function1(Base* _this){
Derived *this = (Derived*)_this;
printf("Derived from Base\n");
}
void Derived_function4(Derived* this){
printf("Only in derived\n");
}
struct Derived_vtable_layout Derived_vtable =
{
{ &Derived_function1,
&Base_function2},
&Derived_function4
};
void Derived_Derived(Derived* this)
{
Base_Base((Base*)this);
this->base.vtable_ptr = (struct Base_vtable_layout*)&Derived_vtable;
}
Derived* new_Derived(){
Derived *res = (Derived*)malloc(sizeof(Derived));
Derived_Derived(res);
return res;
}
int main(){
Derived *der_ptr = new_Derived();
Base *b_ptr = &der_ptr->base;
/* b_ptr->vtable_ptr->function4(b_ptr); Will Give Compilation ERROR!! */
b_ptr->vtable_ptr->function1(b_ptr);
return 0;
}

Q1 - name resolution is static. Since b_ptr is of type Base*, the compiler can't see any of the names unique to Derived in order to access their entries in the v_table.
Q2 - Maybe, maybe not. You have to remember that the vtable itself is simply a very common method of implementing runtime polymorphism and is actually not part of the standard anywhere. No definitive statement can be made about where it resides. The vtable could actually be some static table somewhere in the program that is pointed to from within the object description of instances.
Q3 - If there's a virtual entry in one place there must be in all places otherwise a bunch of difficult/impossible checks would be necessary to provide override capability. If the compiler KNOWS that you have a Base and are calling an overridden function though, it is not required to access the vtable but could simply use the function directly; it can even inline it if it wants.

A1. The vtable pointer is pointing to a Derived vtable, but the compiler doesn't know that. You told it to treat it as a Base pointer, so it can only call methods that are valid for the Base class, no matter what the pointer points to.
A2. The vtable layout is not specified by the standard, it isn't even officially part of the class. It's just the 99.99% most common implementation method. The vtable isn't part of the object layout, but there's a pointer to the vtable that's a hidden member of the object. It will always be in the same relative location in the object so that the compiler can always generate code to access it, no matter which class pointer it has. Things get more complicated with multiple inheritance, but lets not go there yet.
A3. Vtables exist once per class, not once per object. The compiler needs to generate one even if it never gets used, because it doesn't know that ahead of time.

b_ptr points to the Derived vtable- but the compiler can't guarantee that the derived class has a function_4 in it, as that's not contained within the base vtable, so the compiler doesn't know how to make the call and throws an error.
No, the vtable is a static constant somewhere else in the program. The base class merely holds a pointer to it. A Derived class may hold two vtable pointers, but it might not.
In the context of these two classes, then Base needs a vtable to find Derived's function1, which actually is virtual even though you didn't mark it as so, because it overrode a base class virtual function. However, even if this wasn't the case, I'm pretty sure that the compiler is required to produce the vtables anyway, as it has no idea what other classes you have in other translation units that may or may not inherit from these classes and override their virtual functions in unknowable ways.

First, and most important, remember that C++ doesn't do a lot of run-time introspection of any kind. Basically, it needs to know everything about the objects at compile time.
Q1 - b_ptr is a pointer to a Base. Therefore it can only access things that are present in a Base object. No exceptions. Now, the actual implementation may change depending on the actual type of the object, but there's no getting around the method having to be defined in Base if you want to call it through a Base pointer.
Q2 - The simple answer is 'yes, the vtable for a base has to be present in a derived', but there are a LOT of possible strategies for how to layout a vtable, so don't get hung up on it's exact structure.
Q3 - Yes, there must be a vtable in the Base class. Everything that calls virtual functions in a class will go through the vtable so that if the underlying object is actually a Derived, everything can work.
Now that's not an absolute, because if the compiler can be ABSOLUTELY sure that it knows what it's got (as might be the case of a Base object that's declared on the local stack), then the compiler is allowed to optimize out the vtable lookups, and might even be allowed to inline the function.

All of this depends on the implementation. But here are the answers for the usual simplest way using "vtables".
The Base class has a vtable pointer, so the underlying representation is something like this pseudo-code:
struct Base {
void** __vtable;
int a_number;
};
void* __Base_vtable[] = { &Base::function1, &Base::function2 };
void __Base_ctor( struct Base* this_ptr ) { this_ptr->__vtable = __Base_vtable; }
The Derived class includes a Base class subobject. Since that has a place for a vtable, Derived doesn't need to add another one.
struct Derived {
struct Base __base;
};
void* __Derived_vtable[] =
{ &Derived::function1, &Base::function2, &Derived::function4 };
void __Derived_ctor( struct Derived* this_ptr ) {
__Base_ctor( &this_ptr->__base );
this_ptr->__base.__vtable = __Derived_vtable;
}
The "vtable for the Base class", __Base_vtable in my pseudocode, is needed in case somebody tries new Base(); or Base obj;.
All of the above gets more complicated when multiple inheritance or virtual inheritance is involved....
For the line b_ptr -> function4();, this is a compile-time error, not much related to vtables. When you cast to a Base* pointer, you may only use that pointer in the ways defined by class Base (because the compiler doesn't "know" any more whether it's really a Derived, a Base, or some other class). If Derived has a data member of its own, you can't access it through that pointer. If Derived has a member function of its own, virtual or not, you can't access it through that pointer.

Related

Why does a base pointer can access derived member variable in virtual funtion

class Base {
public:
virtual void test() {};
virtual int get() {return 123;}
private:
int bob = 0;
};
class Derived: public Base{
public:
virtual void test() { alex++; }
virtual int get() { return alex;}
private:
int alex = 0;
};
Base* b = new Derived();
b->test();
When test and get are called, the implicit this pointer is passed in. Is it because Derived classes having a sub memory layout that is identical to what a pure base object would be, then this pointer works for both as a base pointer and derived pointer?
Another way to put it is, the memory layout for Derived is like
vptr <-- this
bob
alex
That is why it can use alex in b->test(), right?
Inside of Derived's methods, the implicit this pointer is always a Derived* pointer (more generically, the this pointer always matches the class type being called). That is why Derived::test() and Derived::get() can access the Derived::alex member. That has nothing to do with Base.
The memory layout of a Derived object begins with the data members of Base, followed by optional padding, followed by the data members of Derived. That allows you to use a Derived object wherever a Base object is expected. When you pass a Derived* pointer to a Base* pointer, or a Derived& reference to a Base& reference, the compiler will adjust the pointer/reference accordingly at compile-time to point at the Base portion of the Derived object.
When you call b->test() at runtime, where b is a Base* pointer, the compiler knows test() is virtual and will generate code that accesses the appropriate slot in b's vtable and call the method being pointed at. But, the compiler doesn't know what derived object type b is actually pointing at in runtime (that is the whole magic of polymorphism), so it can't automatically adjust the implicit this pointer to the correct derived pointer type at compile-time.
In the case where b is pointing at a Derived object, b's vtable is pointing at Derived's vtable. The compiler knows the exact offset of the start of Derived from the start of Base. So, the slot for test() in Derived's vtable will point to a private stub generated by the compiler to adjust the implicit Base *this pointer into a Derived *this pointer before then jumping into the actual implementation code for Derived::test().
Behind the scenes, it is roughly (not exactly) implemented like the following pseudo-code:
void Derived_test_stub(Base *this)
{
Derived *adjusted_this = reinterpret_cast<Derived*>(reinterpret_cast<uintptr_t>(this) + offset_from_Base_to_Derived);
Derived::test(adjusted_this);
}
int Derived_get_stub(Base *this)
{
Derived *adjusted_this = reinterpret_cast<Derived*>(reinterpret_cast<uintptr_t>(this) + offset_from_Base_to_Derived);
return Derived::get(adjusted_this);
}
struct vtable_Base
{
void* funcs[2] = {&Base::test, &Base::get};
};
struct vtable_Derived
{
void* funcs[2] = {&Derived_test_stub, &Derived_get_stub};
};
Base::Base()
{
this->vtable = &vtable_Base;
bob = 0;
}
Derived::Derived() : Base()
{
Base::vtable = &vtable_Derived;
this->vtable = &vtable_Derived;
alex = 0;
}
...
Base *b = new Derived;
//b->test(); // calls Derived::test()...
typedef void (*test_type)(Base*);
static_cast<test_type>(b->vtable[0])(b); // calls Derived_test_stub()...
//int i = b->get(); // calls Derived::get()...
typedef int (*get_type)(Base*);
int i = static_cast<get_type>(b->vtable[1])(b); // calls Derived_get_stub()...
The actual details are a bit more involved, but that should give you a basic idea of how polymorphism is able to dispatch virtual methods at runtime.
What you've shown is reasonably accurate, at least for a typical implementation. It's not guaranteed to be precisely as you've shown it (e.g., the compiler might easily insert some padding between bob and alex, but either way it "knows" that alex is at some predefined offset from this, so it can take a pointer to Base, calculate the correct offset from it, and use what's there.
Not what you asked about, so I won't try to get into detail, but just a fair warning: computing such offsets can/does get a bit more complex when/if multiple inheritance gets involved. Not so much for accessing a member of the most derived class, but if you access a member of a base class, it has to basically compute an offset to the beginning of that base class, then add an offset to get to the correct offset within that base class.
A derived class is not a seperate class but an extension. If something is allocated as derived then a pointer (which is just an address in memory) will be able to find everything from the derived class. Classes don't exist in assembly, the compiler keeps track of everything according to how it is allocated in memory and provides appropriate checking accordingly.

What is the underlying mechanism of a base class pointer assigned to derived class

Similar questions I found were more based on what this does; I understand the assignment of a base class pointer to a derived class, e.g Base* obj = new Derived() to be that the right side gets upcasted to a Base* type, but I would like to understand the mechanism for how this happens and how it allows for virtual to access derived class methods. From searching online, someone equated the above code to Base* obj = new (Base*)Derived, which is what led to this confusion. If this type-casting is going on at compile-time, why and how can virtual functions access the correct functions (the functions of the Derived class)? Further, if this casting happens in the way I read it, why do we get errors when we assign a non-inheriting class to Base* obj? Thanks, and apologies for the simplicity of the question. I'd like to understand what causes this behavior.
Note: for clarity, in my example, Derived publicly inherits from Base.
In a strict sense, the answer to "how does inheritance work at runtime?" is "however the compiler-writer designed it". I.e., the language specification only describes the behavior to be achieved, not the mechanism to achieve it.
In that light, the following should be seen as analogy. Compilers will do something analogous to the following:
Given a class Base:
class Base
{
int a;
int b;
public:
Base()
: a(5),
b(3)
{ }
virtual void foo() {}
virtual void bar() {}
};
The compiler will define two structures: one we'll call the "storage layout" -- this defines the relative locations of member variables and other book-keeping info for an object of the class; the second structure is the "virtual dispatch table" (or vtable). This is a structure of pointers to the implementations of the virtual methods for the class.
This figure gives an object of type Base
Now lets look as the equivalent structure for a derived class, Derived:
class Derived : public Base
{
int c;
public:
Derived()
: Base(),
c(4)
{ }
virtual void bar() //Override
{
c = a*5 + b*3;
}
};
For an object of type Derived, we have a similar structure:
The important observation is that the in-memory representation of both the member-variable storage and the vtable entries, for members a and b, and methods foo and bar, are identical between the base class and subclass. So a pointer of type Base * that happens to point to an object of type Derived will still implement an access to the variable a as a reference to the first storage offset after the vtable pointer. Likewise, calling ptr->bar() passes control to the method in the second slot of the vtable. If the object is of type Base, this is Base::bar(); if the object is of type Derived, this is Derived::bar().
In this analogy, the this pointer points to the member storage block. Hence, the implementation of Derived::bar() can access the member variable c by fetching the 3rd storage slot after the vtable pointer, relative to this. Note that this storage slot exists whenever Derived::bar() sits in the second vtable slot...i.e., when the object really is of type Derived.
A brief aside on the debugging insanity that can arise from corrupting the vtable pointer for compilers that use a literal vtable pointer at offset 0 from this:
#include <iostream>
class A
{
public:
virtual void foo()
{
std::cout << "A::foo()" << std::endl;
}
};
class B
{
public:
virtual void bar()
{
std::cout << "B::bar()" << std::endl;
}
};
int main(int argc, char *argv[])
{
A *a = new A();
B *b = new B();
std::cout << "A: ";
a->foo();
std::cout << "B: ";
b->bar();
//Frankenobject
*((void **)a) = *((void **)b); //Overwrite a's vtable ptr with b's.
std::cout << "Franken-AB: ";
a->foo();
}
Yields:
$ ./a.out
A: A::foo()
B: B::bar()
Franken-AB: B::bar()
$ g++ --version
g++ (Ubuntu 5.4.0-6ubuntu1~16.04.5) 5.4.0 20160609
...note the lack of an inheritance relationship between A and B... :scream:
Whoever says
Base* obj = new Derived();
is equivalent to
Base* obj = new (Base*)Derived;
is ignorant of the subject matter.
It's more like:
Derived* temp = new Derived;
Base* obj = temp;
The explicit cast is not necessary. The language permits a derived class pointer to be assigned to a base class pointer.
Most of the time the numerical value of the two pointers are same but they are not same when multiple inheritance or virtual inheritance is involved.
It's the compiler's responsibility to make sure that numerical value of the pointer is offset properly when converting a derived class pointer to a base class pointer. The compiler is able to do that since it makes the decision about the layout of the derived class and the base class sub-objects in the derived class object.
If this type-casting is going on at compile-time, why and how can virtual functions access the correct functions
There is no type casting. There is a type conversion. Regarding the virtual functions, please see How are virtual functions and vtable implemented?.
Further, if this casting happens in the way I read it, why do we get errors when we assign a non-inheriting class to Base* obj?
This is moot since it does not happen the way you thought they did.

Virtual function hiding in derived class

I have two classes related by inheritance:-
class Base
{
public:
virtual void f(int x)
{
cout << "BASE::int" << endl;
}
virtual void f(double x)
{
cout << "BASE::double" << endl;
}
};
class Derived : public Base
{
public:
virtual void f(str::string s)
{
cout << "DERIVED::string" << endl;
}
};
I have provided same method in derived class with different parameters. That means rather than overriding I am hiding base class versions of this function. So, below calls are expected and clear to me.
std::string str("Hello");
Base b;
b.f(1); //calls base class version.
b.f(str); //error.
Derived d;
d.f(1); //error.
d.f(str); //calls derived class version.
But I am not able get clarification for this last scenario.
Base *b = new Derived;
b->f(str); //results in error.
Would compiler not bind this call to derived version of f using vtables and vptrs. But instead it's doing something else. Can anyone provide me complete path how compiler would try to resolve this call as per language mechanisms.
If your pointer is of type Base* then you can only "see" members that are defined in class Base. The compiler doesn't (or pretends not to) "know" that the variable really points to an instance of Derived, even if you just assigned one to it on the previous line.
When you declare a variable to be of type Base*, you're telling the compiler: treat this as something that could point to a Base or to any class derived from it. So you can't access members that are defined in a particular derived class, because there's no guarantee that the pointer actually points to an instance of that derived class.
The vtable only enters the picture at runtime. The generated assembly would have a lookup of the vptr value for a function and a jump to that address. This also means that the polymorphism is "restricted" to functions that Base knows about. Note that this is what makes more sense as well - the definition of a class should only depend on itself and its parents. If you wanted to make Base* b aware of the virtual functions implemented by Derived, you would end up with the number of vtable entries in Bases depending on its children.

query on runtime polymorphism in C++

class base
{
public:
virtual void showbase() {
// ----------
}
};
class base1 {
public:
virtual void showbase1() {
// -------
}
};
class derived : public base, public base1
{
void showbase() {
// ----
}
void showbase1() {
// -------
}
};
int main()
{
base* p = new derived();
p->showbase1();
base1* p1 = new derived();
p1->showbase();
}
As per my understanding about virtual function is that compiler deals it with run time (vtable mechanism), then why I am getting compile time error.
To simulate a compiler, consider what a compiler sees:
class base
{
public:
virtual void showbase() {
// ----------
}
};
base* p = /*blah blah*/;
p->showbase1();
Yes, base is a polymorphic class. And p is indeed a pointer-tobase. But since p points just to a base, and importantly not to a base1 (where showbase1 lives) the compiler interprets the above code like this. Obviously, I'm paraphrasing:
Here is a class named `base` with a single virtual method called `showbase`.
Here is a pointer to a `base` object. Call the method named `showbase1`
And the compiler complains:
Um, excuse me buddy, but base doesn't have a method called
showbase1.
You asked:
[My] understanding about virtual function is that compiler deals with
it at run time. Why I am getting compile time error?
Because the code you've written is nonsense. Here basically is how polymorphism works.
You define a base class with virtual methods.
You define a derived class that overrides those virtual methods.
The compiler creates a vtable which maps the names of the methods in the base class to the implementation in the derived class.
When you call a method in the base class through a pointer (or ref) to the base class, the derived class' implementation is called.
But what you are trying to do is:
Define a base class with virtual methods.
Define a derived class which overrides those virtual methods.
Call a function in a completely different class.
As per my understanding about virtual function is that compiler deals it with run time (vtable mechanism), then why I am getting compile time error.
"Deals with it" is pretty vague and vtables are not magic; In C++ virtual dispatch allows for the actual function called to be one that overrides the statically declared virtual function. That means that the function which is being overridden must be known at compile time.
The vtable does not contain information that would be necessary to look up functions at run-time. Instead, it's basically just a list of pointers to overriding functions. The base provides a complete list of its virtual functions and so, given a particular base type, the compiler knows at compile-time where to go in the vtable for that base for a particular function override; The compiler can generate code that goes directly to that spot in the vtable, gets the pointer, and calls the overriding function.
Then, at run-time, when the actual object of derived type is created, the derived object's constructor fills in the base's vtable, so that anything checking the vtable will get pointers to the derived type's functions.
So the problem with your code is that the function you're calling, showbase(), is not on the list of virtual functions for the type the compiler knows you're accessing, base1; The compiler can't know where in base1's vtable to get a pointer for a function override named showbase(), because there is no such entry in base1's vtable.
A base class pointer to a derived class can only access the member functions defined in the base class. It is illegal to try and access other functions defined in the derived class through it. In your case base class does not define showbase1 and therefore this is illegal
base* p = new derived();
p->showbase1(); //illegal
However, you can do this:
p->showbase(); // legal because showbase is a member function of base
Similarly you can't access showbase1 using a base class pointer
base1* p1 = new derived();
p1->showbase(); //illegal
p1->showbase1(); //legal
Your base class(es) only know about their own member functions, so you can't use it this way. You could do this instead:
base* p = new derived();
p->showbase();
base1* p1 = new derived();
p1->showbase1();
To answer your question about runtime polymorphism, it is dealing with runtime polymorphism (late binding) via the vtable, as you say. But with multiple inheritance, there is essentially a vtable for for each base class. You can't access one base class' vtable via a pointer to the other base class.
p'static type s type is base and hence you can only call with it functions that have been definied into base even if at the end, it will be the functions from derived which will be called because p's dynamic type is derived
Same thing happens for p1.
Maybe you meant p->showbase(); and p1->showbase1();

Would using a virtual destructor make non-virtual functions do v-table lookups?

Just what the topic asks. Also want to know why non of the usual examples of CRTP do not mention a virtual dtor.
EDIT:
Guys, Please post about the CRTP prob as well, thanks.
Only virtual functions require dynamic dispatch (and hence vtable lookups) and not even in all cases. If the compiler is able to determine at compile time what is the final overrider for a method call, it can elide performing the dispatch at runtime. User code can also disable the dynamic dispatch if it so desires:
struct base {
virtual void foo() const { std::cout << "base" << std::endl; }
void bar() const { std::cout << "bar" << std::endl; }
};
struct derived : base {
virtual void foo() const { std::cout << "derived" << std::endl; }
};
void test( base const & b ) {
b.foo(); // requires runtime dispatch, the type of the referred
// object is unknown at compile time.
b.base::foo();// runtime dispatch manually disabled: output will be "base"
b.bar(); // non-virtual, no runtime dispatch
}
int main() {
derived d;
d.foo(); // the type of the object is known, the compiler can substitute
// the call with d.derived::foo()
test( d );
}
On whether you should provide virtual destructors in all cases of inheritance, the answer is no, not necessarily. The virtual destructor is required only if code deletes objects of the derived type held through pointers to the base type. The common rule is that you should
provide a public virtual destructor or a protected non-virtual destructor
The second part of the rule ensures that user code cannot delete your object through a pointer to the base, and this implies that the destructor need not be virtual. The advantage is that if your class does not contain any virtual method, this will not change any of the properties of your class --the memory layout of the class changes when the first virtual method is added-- and you will save the vtable pointer in each instance. From the two reasons, the first being the important one.
struct base1 {};
struct base2 {
virtual ~base2() {}
};
struct base3 {
protected:
~base3() {}
};
typedef base1 base;
struct derived : base { int x; };
struct other { int y; };
int main() {
std::auto_ptr<derived> d( new derived() ); // ok: deleting at the right level
std::auto_ptr<base> b( new derived() ); // error: deleting through a base
// pointer with non-virtual destructor
}
The problem in the last line of main can be resolved in two different ways. If the typedef is changed to base1 then the destructor will correctly be dispatched to the derived object and the code will not cause undefined behavior. The cost is that derived now requires a virtual table and each instance requires a pointer. More importantly, derived is no longer layout compatible with other. The other solution is changing the typedef to base3, in which case the problem is solved by having the compiler yell at that line. The shortcoming is that you cannot delete through pointers to base, the advantage is that the compiler can statically ensure that there will be no undefined behavior.
In the particular case of the CRTP pattern (excuse the redundant pattern), most authors do not even care to make the destructor protected, as the intention is not to hold objects of the derived type by references to the base (templated) type. To be in the safe side, they should mark the destructor as protected, but that is rarely an issue.
Very unlikely indeed. There's nothing in the standard to stop compilers doing whole classes of stupidly inefficient things, but a non-virtual call is still a non-virtual call, regardless of whether the class has virtual functions too. It has to call the version of the function corresponding to the static type, not the dynamic type:
struct Foo {
void foo() { std::cout << "Foo\n"; }
virtual void virtfoo() { std::cout << "Foo\n"; }
};
struct Bar : public Foo {
void foo() { std::cout << "Bar\n"; }
void virtfoo() { std::cout << "Bar\n"; }
};
int main() {
Bar b;
Foo *pf = &b; // static type of *pf is Foo, dynamic type is Bar
pf->foo(); // MUST print "Foo"
pf->virtfoo(); // MUST print "Bar"
}
So there's absolutely no need for the implementation to put non-virtual functions in the vtable, and indeed in the vtable for Bar you'd need two different slots in this example for Foo::foo() and Bar::foo(). That means it would be a special-case use of the vtable even if the implementation wanted to do it. In practice it doesn't want to do it, it wouldn't make sense to do it, don't worry about it.
CRTP base classes really ought to have destructors that are non-virtual and protected.
A virtual destructor is required if the user of the class might take a pointer to the object, cast it to the base class pointer type, then delete it. A virtual destructor means this will work. A protected destructor in the base class stops them trying it (the delete won't compile since there's no accessible destructor). So either one of virtual or protected solves the problem of the user accidentally provoking undefined behavior.
See guideline #4 here, and note that "recently" in this article means nearly 10 years ago:
http://www.gotw.ca/publications/mill18.htm
No user will create a Base<Derived> object of their own, that isn't a Derived object, since that's not what the CRTP base class is for. They just don't need to be able to access the destructor - so you can leave it out of the public interface, or to save a line of code you can leave it public and rely on the user not doing something silly.
The reason it's undesirable for it to be virtual, given that it doesn't need to be, is just that there's no point giving a class virtual functions if it doesn't need them. Some day it might cost something, in terms of object size, code complexity or even (unlikely) speed, so it's a premature pessimization to make things virtual always. The preferred approach among the kind of C++ programmer who uses CRTP, is to be absolutely clear what classes are for, whether they are designed to be base classes at all, and if so whether they are designed to be used as polymorphic bases. CRTP base classes aren't.
The reason that the user has no business casting to the CRTP base class, even if it's public, is that it doesn't really provide a "better" interface. The CRTP base class depends on the derived class, so it's not as if you're switching to a more general interface if you cast Derived* to Base<Derived>*. No other class will ever have Base<Derived> as a base class, unless it also has Derived as a base class. It's just not useful as a polymorphic base, so don't make it one.
The answer to your first question: No. Only calls to virtual functions will cause an indirection via the virtual table at runtime.
The answer to your second question: The Curiously recurring template pattern is commonly implemented using private inheritance. You don't model an 'IS-A' relationship and hence you don't pass around pointers to the base class.
For instance, in
template <class Derived> class Base
{
};
class Derived : Base<Derived>
{
};
You don't have code which takes a Base<Derived>* and then goes on to call delete on it. So you never attempt to delete an object of a derived class through a pointer to the base class. Hence, the destructor doesn't need to be virtual.
Firstly, I think the answer to the OP's question has been answered quite well - that's a solid NO.
But, is it just me going insane or is something going seriously wrong in the community? I felt a bit scared to see so many people suggesting that it's useless/rare to hold a pointer/reference to Base. Some of the popular answers above suggest that we don't model IS-A relationship with CRTP, and I completely disagree with those opinions.
It's widely known that there's no such thing as interface in C++. So to write testable/mockable code, a lot of people use ABC as an "interface". For example, you have a function void MyFunc(Base* ptr) and you can use it this way: MyFunc(ptr_derived). This is the conventional way to model IS-A relationship which requires vtable lookups when you call any virtual functions in MyFunc. So this is pattern one to model IS-A relationship.
In some domain where performance is critical, there exists another way(pattern two) to model IS-A relationship in a testable/mockable manner - via CRTP. And really, performance boost can be impressive(600% in the article) in some cases, see this link. So MyFunc will look like this template<typename Derived> void MyFunc(Base<Derived> *ptr). When you use MyFunc, you do MyFunc(ptr_derived); The compiler is going to generate a copy of code for MyFunc() that matches best with the parameter type ptr_derived - MyFunc(Base<Derived> *ptr). Inside MyFunc, we may well assume some function defined by the interface is called, and pointers are statically cast-ed at compile time(check out the impl() function in the link), there's no overheads for vtable lookups.
Now, can someone please tell me either I am talking insane nonsense or the answers above simply did not consider the second pattern to model IS-A relationship with CRTP?