I am currently developing a library in C++ (Mainly because dependent libraries have C++ interfaces). I created a proof-of-concept implementation with C++ interfaces for quick development. The library has to mandatorily provide C interface for it. The C++ interface is good to have but can be modified/removed if it gets in the way of the C interface.
The C++ API right now looks something like this:
typedef struct {
// only basic C data types like int,float.
int a;
float b;
} Object;
typedef struct {
std::vector<Object> objects;
} GroupOfObjects;
typedef struct {
std::vector<GroupOfObjects> groups;
} Result;
typedef struct {
// some basic C data types
int init1;
int init2;
float init3;
// some C++ types which I can possibly replace with something compatible with C
std::string init4;
std::vector<std::string> init5;
} MyClassInitParams;
struct IMyClass {
public:
virtual bool initialize(MyClassInitParams ¶ms) = 0;
virtual bool getResult(Result &result) = 0;
//Other public methods, constructor, virtual destructor
}
// Actual implementation of the above interface
class MyClass : IMyClass {
}
IMyClass *createMyClassInstance();
I have come up with this C interface till now:
extern "C" {
typedef struct MyClass *MyClassHandle;
// A C version of the above MyClassInitParams
typedef struct{
} MyClassInitParams_C;
typedef struct {
Object * objects;
int numObjects;
} GroupOfObjects_C;
// A C version of the above Result structure
typedef struct{
GroupOfObjects_C *groups;
int numGroups;
}Result_C;
MyClassHandle MyClass_Create();
MyClass_Destroy(MyClassHandle handle);
int MyClass_Initialize(MyClassHandle handle, MyClassInitParams_C *params);
int MyClass_GetResult(MyClassHandle handle , Result_C *result);
void MyClass_FreeResult(Result_C *result);
} // end of extern "C"
Implementation of the C interface:
MyClassHandle MyClass_Create()
{
return createMyClassInstance();
}
MyClass_Destroy(MyClassHandle handle)
{
delete handle;
}
int MyClass_Initialize(MyClassHandle handle, MyClassInitParams_C *params)
{
MyClassInitParam params_cpp;
// fill params_cpp using the params structure
return handle->initialize (params_cpp);
}
int MyClass_GetResult(MyClassHandle handle , Result_C *result)
{
Result result_cpp;
bool ret = handle->getResult(result_cpp);
if (!ret)
return 0;
// Fill the C structure using the cpp structure
result->numGroups = result_cpp.groups.size();
result->groups = new GroupOfObjects_C[result->numGroups];
for (int i = 0; i < result->numGroups; i++) {
result->groups[i].numObjects = result_cpp.groups[i].objects.size();
result->groups[i].objects = new Object[result->groups[i].numObjects];
for (int j = 0; j < result->groups[i].numObjects; j++) {
result->groups[i].objects[j] = result_cpp.groups[i].objects[j];
}
}
return 1;
}
void MyClass_FreeResult(Result_C *result) {
// free all the arrays allocated in the above function
}
I have a few questions regarding this:
The GetResult method has a overhead of copying the objects from C++ vectors to the C arrays. Is there a more elegant and efficient way to handle this?
I will have to maintain the structures for both C and C++. Should I just use the C versions of the MyClassInitParams and Result structures in C++ interface? This will also help with (1).
If I use the solution in (2), does it make sense to even have the C++ interface? Or are there any advantages of keeping both the C and C++ interfaces in this case?
I'd suggest to return Result* from C MyClass_GetResult method, like int MyClass_GetResult(MyClassHandle handle, Result_C **result) or Result_C* MyClass_GetResult(MyClassHandle handle). Then, add accessors for groups and objects.
It's up to you to decide, but I'd rather to use one or the other, but not both.
First of all, I'd suggest to decide what language and its features (C or C++) you're going to use in order to implement your business logic. Next, another language becomes nothing more than a wrapper over logic implemented in another language. Again, if you use functions for accessing actual underlying data you'll get rid of copying this data, as you did in MyClass_GetResult method.
Here's an example
struct Object {
int a;
float b;
};
struct GroupOfObjects;
struct Result;
struct MyClass;
#ifdef __cplusplus
#include <vector>
struct GroupOfObjects {
std::vector<Object> objects;
};
struct Result {
std::vector<GroupOfObjects> groups;
};
struct MyClass {
private:
public:
Result getResult() { /*...*/ }
MyClass(int init1, int init2, float init3, const std::string& init4, const std::vector<std::string>& init5);
}
#endif
#ifdef __cplusplus
extern "C" {
#endif __cplusplus
struct Object* GroupOfObjects_GetObject(struct GroupOfObjects* g, size_t i)
/* { return &g.objects[i]; } */ // commented sections must be in cpp file, not in this header
size_t GroupOfObjects_GetCount(struct GroupOfObjects* g)
/* { return g.objects.size(); } */
struct GroupOfObjects* Result_GetGroup(struct Result* r, size_t i)
/* { return &r.groups[i]; } */
size_t Result_GetGroupCount(struct Result* r)
/* { return g.groups.size(); } */
MyClass *CreateMyClassInstance(int init1, int init2, float init3, const char* init4, const char** init5)
/* {
try {
std::vector<std::string> init5_v;
while (init5 != nullptr)
init5_v.push_back(std::string(*init5++));
return new MyClass(init1, init2, init3, std::string(init4), init5_v);
}
catch (...) {
return nullptr;
}
} */
void FreeMyClassInstance(struct MyClass* mc)
/* { delete mc; } */
Result* MyClass_GetResult(struct MyClass* mc)
/* {
Result *result = nullptr;
try {
result = new Result;
*result = mc->GetResult();
return result;
}
catch (...) {
delete result;
return nullptr;
}
} */
void FreeResult(struct Result* r)
/* { delete r; } */
#ifdef __cplusplus
} // end of extern "C"
#endif
Related
I'm writing a C backend for a C++ library and I want the C code to be able to iterate over the individual items of a forward iterator. In C++ the code which iterates over the items looks like this:
auto rng = wks.range(XLCellReference("A1"), XLCellReference("Q1"));
for (auto& cell : rng) {
// do something with "cell"
}
Now I want to export this code so that it is accessible from C. I'd like to be able to iterate over the items from C using a set of functions like this:
void *startiteration(void *cpp_obj);
void *getnextitem(void *cpp_obj, void *iter);
void finishiteration(void *cpp_obj, void *iter);
I'd imagine startiteration() to return an iterator pointer to the C code which would then be passed along with the C++ object pointer for all successive calls to getnextitem() and finishiteration().
But the problem here is that I can easily pass pointers of objects created in C++ using new between C and C++ code but I don't see how I could do the same with iterators since the iterators are returned by a class method and I don't think there's any way to turn the iterator into a pointer that I could pass to C code which would then pass it back to C++ during the iteration. Also, I don't know how I would "free" such an iterator.
Does anybody have some tips how I can iterate over a C++ forward iterator from C code? How should this be implemented?
EDIT
Implementation based on Silvio's suggestions:
struct myiter
{
XLCellIterator begin;
XLCellIterator end;
};
void *xlsx_startiteration(void *handle, int idx)
{
XLDocument *doc = (XLDocument *) handle;
XLWorkbook wb = doc->workbook();
auto wks = wb.worksheet(wb.sheetNames()[idx-1]);
XLCellRange cr = wks.range(XLCellReference("A1"), XLCellReference("Q1"));
myiter *it = new myiter();
it->begin = std::begin(static_cast<XLCellRange*>(cr));
it->end = std::end(static_cast<XLCellRange*>(cr));
return it;
}
auto rng = wks.range(XLCellReference("A1"), XLCellReference("Q1"));
for (auto& cell : rng) {
// do something with "cell"
}
I like doing 1:1 relationship between C++ and C. The following code outputs 3 lines var=1 var=2 var=3.
By wrapping the objects inside structures, the C side only forward declarations of structures and pointers. The C++ side sees all the rest. Additionally, C side will get a warning when passing invalid pointer to the wrong function.
#include <cstdio>
#include <vector>
typedef int do_not_know_what_is_the_type;
typedef std::vector<do_not_know_what_is_the_type> XLCellRange;
typedef XLCellRange::iterator XLCellIterator;
// header file
// C side
#ifdef __cplusplus
extern "C" {
#endif
struct wks_range_it_s;
void wks_range_it_inc(struct wks_range_it_s *);
bool wks_range_it_ne(struct wks_range_it_s *, struct wks_range_it_s *);
do_not_know_what_is_the_type wks_range_it_deref(struct wks_range_it_s *);
void wks_range_it_destruct(struct wks_range_it_s *);
struct wks_range_s;
struct wks_range_s *wks_range_construct();
void wks_range_destruct(struct wks_range_s *);
struct wks_range_it_s *wks_range_begin(struct wks_range_s *);
struct wks_range_it_s *wks_range_end(struct wks_range_s *);
#ifdef __cplusplus
}
#endif
#ifdef __cplusplus
// C++ side
struct wks_range_s {
XLCellRange rng;
};
struct wks_range_it_s {
XLCellIterator it;
};
extern "C"
void wks_range_it_inc(struct wks_range_it_s *t) {
++t->it;
}
extern "C"
bool wks_range_it_ne(struct wks_range_it_s *a, struct wks_range_it_s *b) {
return a->it != b->it;
}
extern "C"
do_not_know_what_is_the_type wks_range_it_deref(struct wks_range_it_s *t) {
return *t->it;
}
extern "C"
void wks_range_it_destruct(struct wks_range_it_s *t) {
delete t;
}
extern "C"
struct wks_range_s *wks_range_construct() {
// return new struct wks_range_s(wks.range(XLCellReference("A1"), XLCellReference("Q1")));
return new wks_range_s{{1,2,3}};
}
extern "C"
void wks_range_destruct(struct wks_range_s *t) {
delete t;
}
extern "C"
struct wks_range_it_s *wks_range_begin(struct wks_range_s *t) {
return new wks_range_it_s{t->rng.begin()};
}
extern "C"
struct wks_range_it_s *wks_range_end(struct wks_range_s *t) {
return new wks_range_it_s{t->rng.end()};
}
#endif
int main() {
// C code example
struct wks_range_s *rng = wks_range_construct();
// Almost 1:1 relationship with C++ range loop.
for(struct wks_range_it_s *begin = wks_range_begin(rng),
*end = wks_range_end(rng);
wks_range_it_ne(begin, end) ? 1 : (
// Yes, I'm sneaky.
wks_range_it_destruct(begin),
wks_range_it_destruct(end),
0);
wks_range_it_inc(begin)
) {
do_not_know_what_is_the_type var = wks_range_it_deref(begin);
// do something with var here?
printf("var=%d\n", var);
}
wks_range_destruct(rng);
}
Forward iterators in C++ are regular, which means they can be copied. In fact, in our case, we would only need to move them. Specifically, assuming the type of the iterator in question is I, then startiteration could do something like this.
struct Range {
I begin;
I end;
};
void* startiteration(void* cpp_obj) {
Range* my_range = new Range();
my_range->begin = std::begin(*static_cast<MyCppType*>(cpp_object));
my_range->end = std::end(*static_cast<MyCppType*>(cpp_object));
return my_range;
}
Just because the forward iterator is originally returned by value doesn't prohibit us from putting it into a pointer, either directly or indirectly. In our case, we grab both the beginning and the end and store them in a (heap-allocated) structure. Then finishiteration, of course, would be responsible for deleteing the constructed range object.
I have a set of multiple C++ classes that have the same interface (not derived from each other though). I'm trying to wrap these to make them available in .NET.
I currently have a method that defines the wrapper class using C/C++ #defines and then I can subsequently instantiate classes with a simple line of code
However I can't debug this. Ideally I would like to be able to use a generic or a template. However I can't use a C++ type inside a generic which would be the ultimate way to solve this problem.
Has anyone any idea of how I can do this without using the dreaded macros?
EDIT:
OK Here is an example of the templated class I have written:
template< typename CPPResamplerClass >
ref class TResampler
{
CPPResamplerClass* pResampler;
public:
TResampler( int inputSampleRate, int outputSampleRate, int bufferLen ) :
pResampler( new CPPResamplerClass( inputSampleRate, outputSampleRate, bufferLen ) )
{
}
~TResampler()
{
this->!ResamplerName();
}
!TResampler()
{
if (pResampler)
{
delete pResampler;
pResampler = nullptr;
}
}
property int HistorySize
{
int get()
{
return pResampler->HistorySize();
}
}
array< float >^ ResampleAudio(array< float >^ in)
{
pResampler->Get
array< float >^ out = gcnew array< float >(in->Length);
cli::pin_ptr< float > pIn = &in[0];
cli::pin_ptr< float > pOut = &out[0];
unsigned int inLen = in->Length;
unsigned int outLen = out->Length;
if (pResampler->ResampleAudio(pOut, outLen, pIn, inLen))
{
System::Array::Resize(out, outLen);
return out;
}
return nullptr;
}
};
typedef TResampler< ::Vec::SpeexResample > SpeexResample;
I then want to access this from C# however SpeexResample does not exist. This could well be because I am using a typedef ...
Templates don't exist until they're instantiated. While you could instantiate one explicitly:
template ref class TResampler<SomeNativeClass>;
It wouldn't be exactly user-friendly to use from C#. The exported type will literally have angle brackets in its name. Good luck using that. In C# it's only doable through reflection.
The next best thing is to use derived types. Here's a minimal example:
#include "stdafx.h"
#include <iostream>
namespace CppCli {
class NativeClassA
{
int foo;
public:
NativeClassA(int foo) : foo(foo) { std::cout << "Built native class A" << std::endl; }
int getFoo() const { return foo; }
};
class NativeClassB
{
int foo;
public:
NativeClassB(int foo) : foo(foo) { std::cout << "Built native class B" << std::endl; }
int getFoo() const { return foo; }
};
template<typename NativeClass>
public ref class ManagedWrapper
{
NativeClass* ptr;
public:
ManagedWrapper(int foo)
: ptr(new NativeClass(foo))
{}
~ManagedWrapper()
{
this->!ManagedWrapper();
}
!ManagedWrapper()
{
if (ptr)
{
delete ptr;
ptr = nullptr;
}
}
property int Foo { int get() { return ptr->getFoo(); } }
};
public ref class ManagedWrapperA : ManagedWrapper<NativeClassA>
{
public:
ManagedWrapperA(int foo) : ManagedWrapper(foo) {}
};
public ref class ManagedWrapperB : ManagedWrapper<NativeClassB>
{
public:
ManagedWrapperB(int foo) : ManagedWrapper(foo) {}
};
};
Sure enough, ManagedWrapperA and ManagedWrapperB are visible from C#. Maybe you could macro these definitions and still get a decent debugging experience.
I have several CUDA Kernels which are basically doing the same with some variations. What I would like to do is to reduce the amout of code needed. My first thought was to use macros, so my resulting kernels would look like this (simplified):
__global__ void kernelA( ... )
{
INIT(); // macro to initialize variables
// do specific stuff for kernelA
b = a + c;
END(); // macro to write back the result
}
__global__ void kernelB( ... )
{
INIT(); // macro to initialize variables
// do specific stuff for kernelB
b = a - c;
END(); // macro to write back the result
}
...
Since macros are nasty, ugly and evil I am looking for a better and cleaner way. Any suggestions?
(A switch statement would not do the job: In reality, the parts which are the same and the parts which are kernel specific are pretty interweaved. Several switch statements would be needed which would make the code pretty unreadable. Furthermore, function calls would not initialize the needed variables. )
(This question might be answerable for general C++ as well, just replace all 'CUDA kernel' with 'function' and remove '__global__' )
Updated: I was told in the comments, that classes and inheritance don't mix well with CUDA. Therefore only the first part of the answer applies to CUDA, while the others are answer to the more general C++ part of your question.
For CUDA, you will have to use pure functions, "C-style":
struct KernelVars {
int a;
int b;
int c;
};
__device__ void init(KernelVars& vars) {
INIT(); //whatever the actual code is
}
__device__ void end(KernelVars& vars) {
END(); //whatever the actual code is
}
__global__ void KernelA(...) {
KernelVars vars;
init(vars);
b = a + c;
end(vars);
}
This is the answer for general C++, where you would use OOP techniques like constructors and destructors (they are perfectly suited for those init/end pairs), or the template method pattern which can be used with other languages as well:
Using ctor/dtor and templates, "C++ Style":
class KernelBase {
protected:
int a, b, c;
public:
KernelBase() {
INIT(); //replace by the contents of that macro
}
~KernelBase() {
END(); //replace by the contents of that macro
}
virtual void run() = 0;
};
struct KernelAdd : KernelBase {
void run() { b = a + c; }
};
struct KernelSub : KernelBase {
void run() { b = a - c; }
};
template<class K>
void kernel(...)
{
K k;
k.run();
}
void kernelA( ... ) { kernel<KernelAdd>(); }
Using template method pattern, general "OOP style"
class KernelBase {
virtual void do_run() = 0;
protected:
int a, b, c;
public:
void run() { //the template method
INIT();
do_run();
END();
}
};
struct KernelAdd : KernelBase {
void do_run() { b = a + c; }
};
struct KernelSub : KernelBase {
void do_run() { b = a - c; }
};
void kernelA(...)
{
KernelAdd k;
k.run();
}
You can use device functions as "INIT()" and "END()" alternative.
__device__ int init()
{
return threadIdx.x + blockIdx.x * blockDim.x;
}
Another alternative is to use function templates:
#define ADD 1
#define SUB 2
template <int __op__> __global__ void caluclate(float* a, float* b, float* c)
{
// init code ...
switch (__op__)
{
case ADD:
c[id] = a[id] + b[id];
break;
case SUB:
c[id] = a[id] - b[id];
break;
}
// end code ...
}
and invoke them using:
calcualte<ADD><<<...>>>(a, b, c);
The CUDA compiler does the work, build the different function versions and removes the dead code parts for performance optimization.
Consider the following code:
struct data
{
int foo;
int bar;
};
data a;
a.foo = 200;
a.bar = 300;
static void update(data* a, int rspec)
{
if (!rspec) //my data management
{
3rdPartyApi->CreateStream();
3rdPartyApi->PushData(a->foo);
3rdPartyApi->PushData(a->bar);
3rdPartyApi->CloseStream();
}
else // internal data management
{
3rdPartyApi->CreateStream();
3rdPartyApi->PushData(3rdPartyApi->BufferQueue);
3rdPartyApi->CloseStream();
}
3rdPartyApi->PushStream(3rdPartyApi->GetLastStreamBuffer().POD());
}
Lets say I change the value of a.foo or a.bar, and it requires me to call Update there-after the assignment. Can this be done, without actually calling Update() on each change manually?
[EDIT]
Note that the update function created is also assigned to a function pointer for
the third party API, so it can do it's own internal updating. So making the update function non-global is impossible, and thus is why the current update function is global.
[EDIT]
I also rewrote my example to be more understanding and correct to the actual API I'm using
e.g
3rdPartyApi->StreamUpdate((void (*)(void*, int))update);
Yes, you can. Use class methods for this. Pass a static method from your class to the 3rd party API as an update function.
class data
{
public:
void set_foo(int new_foo);
void set_bar(int new_bar);
int get_foo() const;
int get_bar() const;
// This is the update signature which the 3rd party API can accept.
static void update(void* ptr, int rspec);
private:
// These are private so we can control their access.
int foo;
int bar;
};
void data::set_foo(int new_foo)
{
foo = new_foo;
// 'this' is a special pointer for current data object.
update(this);
}
void data::set_bar(int new_bar)
{
bar = new_bar;
update(this);
}
int data::get_foo() const
{
return foo;
}
int data::get_bar() const
{
return bar;
}
// This is needed if the 3rd party API can only call C bindings.
// If it's a C++ API this is not needed.
extern "C" {
void data::update(void* ptr, int rspec)
{
if (!rspec) //my data management
{
// You have to cast to data* from void*.
data* data_ptr = reinterpret_cast<data*>(ptr);
3rdPartyApi->CreateStream();
3rdPartyApi->PushData(data_ptr->foo);
3rdPartyApi->PushData(data_ptr->bar);
3rdPartyApi->CloseStream();
}
else // internal data management
{
3rdPartyApi->CreateStream();
3rdPartyApi->PushData(3rdPartyApi->BufferQueue);
3rdPartyApi->CloseStream();
}
3rdPartyApi->PushStream(3rdPartyApi->GetLastStreamBuffer().POD());
}
} /* extern "C" */
Then:
3rdPartyApi->StreamUpdate(&data::update);
data a;
a.set_foo(200);
a.set_bar(300);
Note that use of a struct instead of a class is equally fine here. But the convention is to use classes in C++. There is only a minor difference which you can learn later.
It is hard to write code for foo, bar, and data, so let's make it more concrete:
class point
{
public:
int x_coord() const;
int y_coord() const;
void move_to(int new_x, int new_y);
private:
void update_3rd_party();
int x;
int y;
};
void point::move_to(int new_x, int new_y)
{
x = new_x;
y = new_y;
// whatever else needs to be done
update_3rd_party();
}
You need to make use of Observer design pattern or a slight variant of it.
See this example here.
The usual way would be to turn foo and bar into some type that overloads the assignment operator:
class updated_int {
int value;
public:
updated_int(int init = 0) : value(init) {}
updated_int &operator=(int new_val) {
value = new_val;
update();
return *this;
}
// You might want to declare this private and not implement it.
updated_int &operator=(updated_int const &r) {
value = r.value;
update();
return *this;
}
operator int() { return value; }
};
struct data {
updated_int foo;
updated_int bar;
}
data a;
a.foo = 1; // operator= will call update() automatically.
I've got some legacy code that, instead of virtual functions, uses a kind field to do dynamic dispatch. It looks something like this:
// Base struct shared by all subtypes
// Plain-old data; can't use virtual functions
struct POD
{
int kind;
int GetFoo();
int GetBar();
int GetBaz();
int GetXyzzy();
};
enum Kind { Kind_Derived1, Kind_Derived2, Kind_Derived3 /* , ... */ };
struct Derived1: POD
{
Derived1(): kind(Kind_Derived1) {}
int GetFoo();
int GetBar();
int GetBaz();
int GetXyzzy();
// ... plus other type-specific data and function members ...
};
struct Derived2: POD
{
Derived2(): kind(Kind_Derived2) {}
int GetFoo();
int GetBar();
int GetBaz();
int GetXyzzy();
// ... plus other type-specific data and function members ...
};
struct Derived3: POD
{
Derived3(): kind(Kind_Derived3) {}
int GetFoo();
int GetBar();
int GetBaz();
int GetXyzzy();
// ... plus other type-specific data and function members ...
};
// ... and so on for other derived classes ...
and then the POD class's function members are implemented like this:
int POD::GetFoo()
{
// Call kind-specific function
switch (kind)
{
case Kind_Derived1:
{
Derived1 *pDerived1 = static_cast<Derived1*>(this);
return pDerived1->GetFoo();
}
case Kind_Derived2:
{
Derived2 *pDerived2 = static_cast<Derived2*>(this);
return pDerived2->GetFoo();
}
case Kind_Derived3:
{
Derived3 *pDerived3 = static_cast<Derived3*>(this);
return pDerived3->GetFoo();
}
// ... and so on for other derived classes ...
default:
throw UnknownKindException(kind, "GetFoo");
}
}
POD::GetBar(), POD::GetBaz(), POD::GetXyzzy(), and other members are implemented similarly.
This example is simplified. The actual code has about a dozen different subtypes of POD, and a couple dozen methods. New subtypes of POD and new methods are added pretty frequently, and so every time we do that, we have to update all these switch statements.
The typical way to handle this would be to declare the function members virtual in the POD class, but we can't do that because the objects reside in shared memory. There is a lot of code that depends on these structs being plain-old-data, so even if I could figure out some way to have virtual functions in shared-memory objects, I wouldn't want to do that.
So, I'm looking for suggestions as to the best way to clean this up so that all the knowledge of how to call the subtype methods is centralized in one place, rather than scattered among a couple dozen switch statements in a couple dozen functions.
What occurs to me is that I can create some sort of adapter class that wraps a POD and uses templates to minimize the redundancy. But before I start down that path, I'd like to know how others have dealt with this.
You can use a jump table. This is what most virtual dispatches look like under the hood, and you CAN construct it manually.
template<typename T> int get_derived_foo(POD*ptr) {
return static_cast<T>(ptr)->GetFoo();
}
int (*)(POD*) funcs[] = {
get_derived_foo<Derived1>,
get_derived_foo<Derived2>,
get_derived_foo<Derived3>
};
int POD::GetFoo() {
return funcs[kind](this);
}
For a short example.
What exactly are the limitations of being in shared memory? I realized that I don't know enough here. Does it mean that I can't use pointers, because someone in another process will be trying to use those pointers?
You could use a string map, where each process gets it's own copy of the map. You'd have to pass this in to GetFoo() so that it can find it.
struct POD {
int GetFoo(std::map<int, std::function<int()>& ref) {
return ref[kind]();
}
};
Edit: Of course, you don't have to use a string here, you could use an int. I just used it as example. I should change it back. Infact, this solution is pretty flexible, but the important thing is, make a copy of process-specific data, e.g. function pointers or whatever, and then pass it in.
Here's the template-metaprogramming path I'm going down now. Here is what I like about it:
Adding support for a new kind only requires updating LAST_KIND and adding a new KindTraits.
There is a simple pattern for adding a new function.
Functions can be specialized for particular kinds if necessary.
I can expect compile-time errors and warnings, rather than mysterious run-time misbehavior, if I screw anything up.
There are a couple of concerns:
POD's implementation is now dependent upon the interfaces of all the derived classes. (This is already true in the existing implementation, so I'm not worried about it, but it is a bit of a smell.)
I'm counting on the compiler to be smart enough to generate code that is roughly equivalent to the switch-based code.
Many C++ programmers will scratch their heads upon seeing this.
Here's the code:
// Declare first and last kinds
const int FIRST_KIND = Kind_Derived1;
const int LAST_KIND = Kind_Derived3;
// Provide a compile-time mapping from a kind code to a subtype
template <int KIND>
struct KindTraits
{
typedef void Subtype;
};
template <> KindTraits<Kind_Derived1> { typedef Derived1 Subtype; };
template <> KindTraits<Kind_Derived2> { typedef Derived2 Subtype; };
template <> KindTraits<Kind_Derived3> { typedef Derived3 Subtype; };
// If kind matches, then do the appropriate typecast and return result;
// otherwise, try the next kind.
template <int KIND>
int GetFooForKind(POD *pod)
{
if (pod->kind == KIND)
return static_cast<KindTraits<KIND>::Subtype>(pod)->GetFoo();
else
return GetFooForKind<KIND + 1>(); // try the next kind
}
// Specialization for LAST_KIND+1
template <> int GetFooForKind<LAST_KIND + 1>(POD *pod)
{
// kind didn't match anything in FIRST_KIND..LAST_KIND
throw UnknownKindException(kind, "GetFoo");
}
// Now POD's function members can be implemented like this:
int POD::GetFoo()
{
return GetFooForKind<FIRST_KIND>(this);
}
You can experiment with Curiously recurring template pattern. It's a bit complicated, but when you cannot use pure virtual functions it can be helpful.
Here is an approach that uses virtual methods to implement the jump table, without requiring the Pod class or the derived classes to actually have virtual functions.
The objective is to simplify adding and removing methods across many classes.
To add a method, it needs to be added to Pod using a clear and common pattern, a pure virtual function needs to be added to PodInterface, and a forwarding function must be added to PodFuncs using a clear and common pattern.
Derived classes need only have a file static initialisation object to set things up, otherwise look pretty much like they already do.
// Pod header
#include <boost/shared_ptr.hpp>
enum Kind { Kind_Derived1, Kind_Derived2, Kind_Derived3 /* , ... */ };
struct Pod
{
int kind;
int GetFoo();
int GetBar();
int GetBaz();
};
struct PodInterface
{
virtual ~PodInterface();
virtual int GetFoo(Pod* p) const = 0;
virtual int GetBar(Pod* p) const = 0;
virtual int GetBaz(Pod* p) const = 0;
static void
do_init(
boost::shared_ptr<PodInterface const> const& p,
int kind);
};
template<class T> struct PodFuncs : public PodInterface
{
struct Init
{
Init(int kind)
{
boost::shared_ptr<PodInterface> t(new PodFuncs);
PodInterface::do_init(t, kind);
}
};
~PodFuncs() { }
int GetFoo(Pod* p) const { return static_cast<T*>(p)->GetFoo(); }
int GetBar(Pod* p) const { return static_cast<T*>(p)->GetBar(); }
int GetBaz(Pod* p) const { return static_cast<T*>(p)->GetBaz(); }
};
//
// Pod Implementation
//
#include <map>
typedef std::map<int, boost::shared_ptr<PodInterface const> > FuncMap;
static FuncMap& get_funcmap()
{
// Replace with other approach for static initialisation order as appropriate.
static FuncMap s_funcmap;
return s_funcmap;
}
//
// struct Pod methods
//
int Pod::GetFoo()
{
return get_funcmap()[kind]->GetFoo(this);
}
//
// struct PodInterface methods, in same file as s_funcs
//
PodInterface::~PodInterface()
{
}
void
PodInterface::do_init(
boost::shared_ptr<PodInterface const> const& p,
int kind)
{
// Could do checking for duplicates here.
get_funcmap()[kind] = p;
}
//
// Derived1
//
struct Derived1 : Pod
{
Derived1() { kind = Kind_Derived1; }
int GetFoo();
int GetBar();
int GetBaz();
// Whatever else.
};
//
// Derived1 implementation
//
static const PodFuncs<Derived1>::Init s_interface_init(Kind_Derived1);
int Derived1::GetFoo() { /* Implement */ }
int Derived1::GetBar() { /* Implement */ }
int Derived1::GetBaz() { /* Implement */ }
Here is an example using Curiously recurring template pattern. This may suit your needs if you know more info at the compile time.
template<class DerivedType>
struct POD
{
int GetFoo()
{
return static_cast<DerivedType*>(this)->GetFoo();
}
int GetBar()
{
return static_cast<DerivedType*>(this).GetBar();
}
int GetBaz()
{
return static_cast<DerivedType*>(this).GetBaz();
}
int GetXyzzy()
{
return static_cast<DerivedType*>(this).GetXyzzy();
}
};
struct Derived1 : public POD<Derived1>
{
int GetFoo()
{
return 1;
}
//define all implementations
};
struct Derived2 : public POD<Derived2>
{
//define all implementations
};
int main()
{
Derived1 d1;
cout << d1.GetFoo() << endl;
POD<Derived1> *p = new Derived1;
cout << p->GetFoo() << endl;
return 0;
}
Expanding on the solution you ended up with, the following solves the mapping to derived functions at program initialization:
#include <typeinfo>
#include <iostream>
#include <functional>
#include <vector>
enum Kind
{
Kind_First,
Kind_Derived1 = Kind_First,
Kind_Derived2,
Kind_Total
};
struct POD
{
size_t kind;
int GetFoo();
int GetBar();
};
struct VTable
{
std::function<int(POD*)> GetFoo;
std::function<int(POD*)> GetBar;
};
template<int KIND>
struct KindTraits
{
typedef POD KindType;
};
template<int KIND>
void InitRegistry(std::vector<VTable> &t)
{
typedef typename KindTraits<KIND>::KindType KindType;
size_t i = KIND;
t[i].GetFoo = [](POD *p) -> int {
return static_cast<KindType*>(p)->GetFoo();
};
t[i].GetBar = [](POD *p) -> int {
return static_cast<KindType*>(p)->GetBar();
};
InitRegistry<KIND+1>(t);
}
template<>
void InitRegistry<Kind_Total>(std::vector<VTable> &t)
{
}
struct Registry
{
std::vector<VTable> table;
Registry()
{
table.resize(Kind_Total);
InitRegistry<Kind_First>(table);
}
};
Registry reg;
int POD::GetFoo() { return reg.table[kind].GetFoo(this); }
int POD::GetBar() { return reg.table[kind].GetBar(this); }
struct Derived1 : POD
{
Derived1() { kind = Kind_Derived1; }
int GetFoo() { return 0; }
int GetBar() { return 1; }
};
template<> struct KindTraits<Kind_Derived1> { typedef Derived1 KindType; };
struct Derived2 : POD
{
Derived2() { kind = Kind_Derived2; }
int GetFoo() { return 2; }
int GetBar() { return 3; }
};
template<> struct KindTraits<Kind_Derived2> { typedef Derived2 KindType; };
int main()
{
Derived1 d1;
Derived2 d2;
POD *p;
p = static_cast<POD*>(&d1);
std::cout << p->GetFoo() << '\n';
p = static_cast<POD*>(&d2);
std::cout << p->GetBar() << '\n';
}