Reducing virtually duplicate template instantiations - c++

I have a (generated) set of classes that look roughly like this when simplified:
class A {
public:
static const int id = 0;
typedef int value_type;
void* data; // actually pointer to int
};
class B {
public:
static const int id = 1;
typedef float value_type;
void* data; // actually pointer to float
};
There are altogether tens of these, with much fewer types. I need to create classes that calculate some derived values for all of these. Now the obvious solution would be using a template like this:
template<class T>
class Derived {
public:
typename T::value_type value;
void update(const void* data) {
value = *static_cast<typename T::value_type*>(data);
// do some calculations
}
};
But this is going to instantiate a separate class for every single parameter class A, B, and so on; most of which will be identical. The other solution would be an obvious template class like this:
template<typename T>
class Derived2 {
public:
T value;
void update(const void* data) {
value = *static_cast<T*>(data);
}
};
This would be updating the code using this class manually if the schema that classes A, B, etc., are generated from changes. Is there a way to use the value_type typedefs to generate instantiations of Derived2<int>, Derived2<float> etc., or at least match the types of manually parametrized instantiations to types A, B, etc.?
This is an embedded system, so the target is reducing the amount of identical code, even if it leads to more convoluted C++ code. Replacing the void*s with actual types would lead to code explosion in other parts of the program, so it is not done.

As posted in the comments this will do what you need:
template<class T> struct Derived : Derived2<typename T::value_type> { };

Related

c++ defining type of a member class without template argument

I am trying to set the type of the member of a class, without passing it through template argument.
In details:
// Forward declaration:
class A;
class B;
class Base
{
};
template <class T>
class Derived : public Base
{
private:
T2 var;
};
where T could be either class A or class B.
What I would like to do is for Derived<A> T2 is int (for instance) and for Derived<B> T2 is double (for instance). I would like to avoid the following solution:
template <class T1, class T2>
class Derived : public Base
{
private:
T2 var;
};
I want to avoid this because for Derived<A> there could be various possible combinations for T2: Derived<A,int>, Derived<A,double>, ...
What I want is that the type of T2 is unique for the entire Derived<A>.
Any idea how to solve that ?
Update: The comments show that the original problem you are trying to solve is not completely explained in the question. I'll leave the original answer nevertheless at the bottom of this answer.
You cannot have two Derived<A> with different types T2 for the var member. In addition, a variable defined by the User can not influence the type of the member variable. Variable values are set at runtime, types are determined at compiletime.
To store a type somehow defined by the user, you will have either have to restrict the variable to a set of known types or use one type that contains a serialized version of the variable's content. The set of known types is often used in the context of databases, where the fields can have one of several predefined types (e.g. String, Integer, Boolean, Double). The type for the member variable then could be a Boost.Variant, restricted to C++ representations of that type. Another application of "user defined types" are where the user of your program has to somehow define the layout and interpretation of the type and its object, for example if your program is the interpreter of some scripting language. In that case again a Boost.Variant (or something similar) can be of use, or, since the value is probably some user provided input, just store the serialized value in a string and interpret it every time you have to deal with it.
Original answer:
This is usually done via template metaprogramming, in this case a type function (sometimes, depending on the context, part of a traits or policy class):
template <class T>
struct DerivedMemVarType {
typedef double type; //default
};
template<>
struct DerivedMemVarType<A> {
typedef int type;
};
And then:
template <class T>
class Derived : public Base
{
typedef typename DerivedMemVarType<T>::type T2;
private:
T2 var;
};
You can also leave out the default, so that any instantiation of Derived for a type that you have not mapped in your function will give a compile error:
template <class T>
struct DerivedMemVarType; //default is not defined
template<>
struct DerivedMemVarType<A> {
typedef int type;
};
template<>
struct DerivedMemVarType<B> {
typedef double type;
};
//...
Derived<C> dc; // ...template error mess....
// --> error: invalid use of incomplete type 'struct DerivedMemVarType<C>'
if you do not have any type specific function call, you can use something like...
class A;
class B;
class Base
{
};
template <class T>
class Derived : public Base
{
public:
Derived(T startVal):var(startVal){}
private:
T var;
};
template <typename T>
Derived<T> MakeDerived(T initValue)
{
return Derived<T>(initValue);
}
and now you can use it the following and the compiler should know what type you are passing to the function.
int initialValue = 0;
auto derived = MakeDerived(initialValue);
I think you can create a separate class that just holds a typedef which you then specialize and use in your Derived class.
template<typename T>
class VarType {
public:
typedef int TheType;
}
template <>
class VarType<B> {
public:
typedef double TheType;
};
template <typename T>
class Derived : public Base {
private:
typename VarType<T>::TheType var;
};

vector of template struct

using namespace std;
#include <vector>
#include <string>
template <class T>
struct ValNode {
string id;
T value;
};
class ValTable {
public:
ValTable();
template <class T>
void add(string,T);
const bool find(string);
void remove(string);
private:
template<class T>
std::vector<ValNode<T>*> vals;
};
complier error:error: data member 'vals' cannot be a member template
i did try to use T* value in the struct, but i didnt work out.
I didnt use any of the functions in codes yet. was just trying to compling it into *.o file (with .cpp file also).
As the error says, variables (including data members) can't be templates; only classes and functions can be.
It looks like you want the table to be able to hold values of various different types, specified at run-time according to which types are passed to add(). For that, you need dynamic types, which aren't directly supported in C++. You might consider libraries like Boost.Any or Boost.Variant for that.
On the other hand, maybe you just want to store a single type in each table, and different types in different tables. In that case, the class itself will need to be a template:
template <typename T>
class ValTable {
public:
ValTable();
void add(string,T);
const bool find(string);
void remove(string);
private:
std::vector<ValNode<T>*> vals;
};
In C++ you can have template methods in a class, but not template data members.
For example:
template<typename T, int n>
struct FixedVector {
T x[n];
FixedVector() {
for (int i=0; i<n; i++) x[i] = 0;
}
template<typename C>
void copy(const C& container) {
if (container.size() != n) {
throw std::runtime_error("Wrong size");
}
int j = 0;
for (typename C::const_iterator i=container.begin(),
e=container.end();
i!=e;
++i)
{
x[j++] = *i;
}
}
};
With the above class you can declare FixedVector<int, 5> f and call f.copy(v) where v can be for example a vector or a list or anything that has size begin and end.
So FixedVector::copy is a template method and this means that the compiler will generate a different version of it for each different type you will pass to the function.
std::vector<double> y;
y.push_back(3.4); y.push_back(5.6); y.push_back(7.8);
std::list<unsigned char> z;
z.push_back('a'); z.push_back('b'); z.push_back('c');
FixedVector<int, 3> v;
v.copy(y); // This is ok
v.copy(z); // This is ok too
C++ doesn't allow template data members because this would imply a different class size depending on how many types you are using in a specific compilation unit and this doesn't go with the C++ compilation model of one-unit-at-a-time.
Adding methods is instead fine because it doesn't affect class size, and everything can be fixed at link time by avoiding pulling multiple copies of the same method from different compilation units.
You will have to declare ValTable as template
template <class T>
class ValTable{
public:
ValTable();
//template <class T>
void add(string,T);
const bool find(string);
void remove(string);
private:
//template<class T>
std::vector<ValNode<T>*> vals;
};
You won't be able to do that ValTable needs to be a template
You can have this :
template <class T> //Make the class as template
class ValTable {
public:
ValTable();
template <class X>
void add(string,X);
const bool find(string);
void remove(string);
private:
//template<class T>
std::vector<ValNode<T>*> vals;
};
You cannot have template member values: each translation unit could access different instantiations resulting in different ibject layouts. You'd need to factor out the type in some way.
The standard library does something along the lines of what you want for std::locale: each std::locale stores a collection of differently typed objects. It is soecial purpose and can't be used for your purpose directly.
The basic idea is to automatically map each type used to an int which is then used to map the type to an instance. The vals member would then be a function template looking up the correct instance. A rough outline could look like this:
int type_index_alloc() {
static std::atomic<int> current(0);
return ++current;
}
template <typename T>
int type_map() {
static int rc = type_index_alloc();
}
class container {
struct base { virtual ~base() {} };
template <typename T>
struct value: base { T v; };
std::map<int, std::shared_ptr<base>> vals_;
public:
T& vals() {
std::shared_ptr<base>& rc(vals_[type_map<T>()]);
if (!rc) {
rc = std::make_shared<value<T>>()); }
return static_cast<value<T>&>(*rc).v;
}
};
This a just trying to show how things are being set up: I currently don't have access to a compiler. Also, the code example just provides access to an object of type T but it could easily be changed to use a std::vector<T> instead.

How to create an array of templated class objects?

I haven't done any C++ programming for quite some time and I decided that I would mess around with it a little bit in my spare time so I decided to write me a little database program just for fun and I'm having trouble with creating an array of templated class objects.
What I have is this class which I want to use to represent a field in a database record.
template <class T, int fieldTypeId>
class Field
{
private:
T field;
int field_type;
public:
// ...
};
And I want to use an array of that class to represent a record in a database using this class.
class Database_Record
{
private:
int id;
Field record[];
public:
Database_Record(int);
Database_Record(int, Field[]);
~Database_Record();
};
Where I'm stuck at is the creation of the array in the Database_Record class since that is an array of templated class objects with each element possibly being of a different type and I'm not sure how I need declare the array because of that. Is what I'm trying to do even possible or am I going about it the wrong way? Any help would be greatly appreciated.
Field<T1> and Field<T2> are two completely different types. To treat them in a vector you need to generialize then somewhere. You may write AbstractField and
struct AbstractField{
virtual ~AbstractField() = 0;
};
template<class T,int fieldTypeId>
class Field: public AbstractField{
private:
T field;
public:
const static int field_type;
public:
virtual ~Field(){}
};
class Database_Record{
std::vector<AbstractField*> record;
public:
~Database_Record(){
//delete all AbstractFields in vector
}
};
and then keep a vector of AbstractField. also use vector instead of []. Use AbstractField* instead of AbstractField and write at least one pure virtual in AbstractField.
you may make the destructor of AbstractField pure virtual. and don't forget to delete all AbstractFields. in ~Database_Record()
You are going the wrong way.
Templates are used to create distinct types: std::vector<int> and std::vector<float> are distinct in much the same way (and as much) as int and float are.
Your syntax is also wrong; to create a dynamic array you'd put the following member in your Database_Record:
std::vector<Field> record; // if this was possible; however, it's not
To put several objects of distinct type into a single array, they ought to have a common base class.
In order to create an array of different types you need a base class for the objects and the array will be an array of pointers to that base class. So, for example,
class Field
{
public:
virtual ~Field() {}
virtual std::string toString() const = 0;
// and possibly other interface functions...
};
template <class T> FieldImpl : public Field
{
public:
virtual std::string toString() const
{
std::stringstream ss;
ss << val;
return ss.str();
}
// implementation of possibly other interface functions
private:
T val;
}
will be the types you need. The array will then be something like
std::vector<std::unique_ptr<Field>> my_array;
You can then do stuff with your array using the interface functions, e. g.
my_array[i]->toString();
As has been said before, C++ templates don't work like that.
At the same, using inheritance and vectors of pointers is not suitable for implementations of DB records because of performance limitations.
Take a step back and look at the problem in a more abstract way. As I understand from your code, the intent is to package an arbitrary number of fields of different types into a continuous memory block. Schematically:
struct DBRecord {
Type1 f1;
Type2 f2;
Type3 f3;
Type4 f4;
// etc...
}
You can achieve this by a bit ugly but practical construct consisting of an abstract template declaration and several specializations.
The declaration would look like this:
template <
typename T1,
typename T2 = void,
typename T3 = void,
typename T4 = void,
typename T5 = void,
typename T6 = void,
typename T7 = void,
typename T8 = void,
typename T9 = void,
typename T10 = void
> struct DBRecord;
It limits a max number of fields to some specific number obviously. If you need a truly arbitrary number of fields you need to switch to column-oriented paradigm.
Then, partial specializations should declare anatomy of structures for each number of arguments from 1 to 10:
template <
typename T1
> struct DBRecord <T1, void, void, void, void, void, void, void, void, void>
{
int id;
T1 f1;
DBRecord(int ID, T1 F1) {/*...*/};
};
template <
typename T1,
typename T2
> struct DBRecord <T1, T2, void, void, void, void, void, void, void, void>
{
int id;
T1 f1;
T2 f2;
DBRecord(int ID, T1 F1, T2 F2) {/*...*/};
};
// etc...
Now, you can allocate tables as arrays of records of certain types in one new[] call if you want.
And, you don't normally care about destruction of each field, since you free memory of the whole structure.
Macros can help to make declaration of such specializations somewhat more compact.
Consider every instantiation with a different template argument to be a different class. You either need to store a specific class (ie Field<int, 17>) or you need Field to have a non-templated base class which you can store in a list.
You can do something like this -
template <class T, int fieldTypeId>
class Field
{
private:
T field;
int field_Type;
};
template <class T, int fieldTypeId>
class Database_record
{
private:
int id;
std::vector<Field<T, fieldTypeId> > record_;
};
You are doing templates wrong. Instantiating class templates with
different types will yield two different types again with possibly
different sizes, which makes it impossible to store them in an array.
If you want to treat different types uniformly, use inheritance. And
when you use inheritance, don't use plain arrays, but vector or
std::array.
There is also a bunch of odd things in your code: Why store a
fieldTypeId when it is known statically? I guess it is related to
the type T you are using as a template parameter. Externalise the
mechanism through partial specializations:
template<typename T>
struct fieldTypeId;
template<>
struct fieldTypeId<int> {
const static int value = 0;
};
// etc....
If I'm completely wrong and you really know what you are doing: Use
type erasure through some any type (e.g. Boost.Any).
Made 2 example classes used for quick debug reports inspired by ToString() override of C#:
class UnknownType_t {
public:
virtual operator long&() { throw "Unsupported"; };
virtual operator const std::string() { throw "Unsupported"; };
virtual void Set(char*, long) = 0;
};
class Number : public UnknownType_t {
public:
Number(long _n) { n = _n; };
virtual operator long&() { return n; };
virtual void Set(char* buf, long size) {
n = 0;
memcpy(&n, buf, size);
}
long n;
};
class String : public UnknownType_t {
public:
String(const char *_s) : s(_s) {};
virtual operator const std::string() { return s; };
virtual void Set(char* buf, long size) {
s = std::string(reinterpret_cast<char*>(buf), size);
}
std::string s;
};
You can check type trying dynamic_cast, result in common array of UnknownType_t look like
{n=123 } or {s="ABC" }.
Base is not pure virtual by intent - required cross getters would have no sense...

How to count the number of CRTP subclasses of a template class?

Does anyone know of a method to use CRTP to count the number of subclasses of an object?
Suppose we had a setup similar to the following one:
template <typename T>
class Object
{
....
};
const unsigned int ObjectSubClassCount = ...;
class Subobject : public Object<SubObject>
{
....
};
class Second : public Object<Second>
{
....
};
and so on, such that, using TMP, we might have a constant (ObjectSubClassCount) that represents the total number of subclasses?
Does anyone know a way to do this?
Edit: I am wanting to use the result as a template parameter later on, so I need it to be done with TMP...
Without the requirement to use the result as a template parameter later I would try it doing like this:
// Class which increments a given counter at instanciation
struct Increment {
Increment(std::size_t& counter)
{
counter++;
}
};
// This is your template base
template <typename T>
class Object
{
private:
// For every instanciation of the template (which is done for a subclass)
// the class counter should get incremented
static Increment incrementCounter;
};
// This is the global object counter
static std::size_t classCounter;
// Static Member Variable
template<typename T>
Object<T>::incrementCounter(classCounter);
Haven't tried it but should do. To have the result available as a template parameter again (MPL) I don't have enough experience in MPL but I doubt this is possible.
Ok, so I've come upon a... somewhat acceptable answer.
I figured that it would not work out if the subclasses had absolutely not knowledge of eachother (I mean, we're talking somewhat functional programming...).
Here's a solution for this. It's definitely not the solution I'd wish for; however, it is a start. I've forced all objects to use a form of CRTP, but one that uses more of a linked list format. In this way, our subclasses must be derived from an Object<> templated from:
A: itself
and B: the most recent previously defined subclass
here is my code for this (I use a template from <type_traits> once, just a note)
template <typename T> //SFINAE check for the existance of subclasscount static member
struct has_subclasscount
{
template <typename U>
static typename std::enable_if<sizeof(U::subclasscount) != 0, int>::type test(int);
template <typename U>
static char test(...);
static const bool result = (sizeof(test<T>(0)) == sizeof(int))?(true):(false);
};
template <bool res, typename T>
struct return_subclasscount //the value to return is 0 if false
{
static const int result = 0;
};
template <typename T>
struct return_subclasscount<true, T> //returns subclasscount only if the first parameter is true
{
static const int result = T::subclasscount;
};
template <typename T> //combines return_subclasscount and has_subclasscount
struct get_subclasscount
{
static const int result = return_subclasscount<has_subclasscount<T>::result, T>::result;
};
template <typename This, typename Prev>
class Object
{
public:
static const int subclasscount = 1 + get_subclasscount<Prev>::result; //the subclass count
};
class sub1 : public Object<sub1, int>
{
};
class sub2 : public Object<sub2, sub1>
{
};
class sub3 : public Object<sub3, sub2>
{
};
These last 3 empty classes are the subclasses that we're counting. This is our header file.
In our main .cpp file, we have:
int main() {
std::cout << sub3::subclasscount;
char c;
std::cin >> c;
}
Running it, we get a simple output of:
3
Which confirms that it has worked.
Now, some of the downsides to this solution is:
We must know what our last defined subclass was, before we add on.
We must keep up with anywhere we use the subclass counter, always modifying it to be from the last subclass in the list (this can be averted by using a consistant "endoflist" subclass, which would need to be maintained instead)
Upsides, though, include the fact that we do not need to upkeep any of our previously defined subclasses. However, I consider this answer more of a "starting point" than a "final solution"; perhaps something that can be expanded upon?
(also, this can easily be abused to make a form of tree structure, where subclasscount would actually represent the depth of any given node in the tree)
Anyone have any ideas from here?

Downside of this macro construct and possible alternatives

I recently saw some code using macros like
#define CONTAINS(Class, Name)\
private:\
std::list<Class> m_##Name##s;\
public:\
void add_##Name(const Class& a_##Name) {\
m_##Name##s.push_back(a_##Name);\
}\
int get_##Name(int pos) {\
return m_##Name##s.at(pos);\
}\
// ... more member functions
Later you can declare a class like
class my_class {
CONTAINS(int, integer)
// ...
};
and write
my_class a(...);
a.add_integer(10);
I was puzzled about this paste-in-macro-style because I'm missing concrete counter-arguments. But beside that I accept the following pros
you can easily add a list interface for arbitrary types to your class
you avoid frequently repeated code
you have an easy to use interface (like add_integer(10))
Now I'm searching for alternatives which meet all these points above and avoid this old C macro style. My first idea was to create an abstract base class template
template<typename T>
class list_interface {
private:
std::list<T> m_list;
public:
void add_element(const T& x) {
m_list.push_back(x);
}
// ... more member functions
};
and add it to my class via inheritance like this
class my_class : public list_interface<int> {
// ...
};
Now I can write too
my_class a;
a.add_element(10);
but I'm concerned about the following points:
you can only add one list to your class
you publicly inherit from a class without virtual destructor
I don't meet the third point of the pros (add_element(10) instead of add_integer(10))
My questions are:
What are the drawbacks of the old C macro construct
How can I provide a similar functionality without macros
How about:
#include <vector>
template<typename T>
class Plop
{
std::vector<T> data;
public:
void add(T const& v) {data.push_back(v);}
T get(int pos) {return data.at(pos);} // at() is not valid on lists.
};
class my_class
{
public:
Plop<int> integer;
Plop<float> floater;
};
int main()
{
my_class x;
x.integer.add(5); // similar to x.add_integer(5);
x.integer.get(0); // similar to x.get_integer(0);
}
It meets all the requirements:
you can easily add a list interface for arbitrary types to your class
you avoid frequently repeated code
you have an easy to use interface (like add_integer(10))
My questions are:
What are the drawbacks of the old C macro construct
The ick factor.
Debugging.
Can pass individual list to other functions or methods.
How can I provide a similar functionality without macros
See above.
my opinion
1) Yuck yuck. Complex macro which will hinder debugging. Just the view of the macro makes my skin crawl.
2) Your inheritance solution looks fine. If you really needed multiple lists of different types, you might want to consider just writing more code and instantiating the list as a member variable. There really is no benefit of attempting to reduce lines of code by making it convoluted.
There is a way, in meta-programming fashion, and using tags.
First, let's consider the roll your own solution.
The idea is to come up with this interface:
class my_class : public vector<Name, std::string>, public vector<Foo, int>
{
};
And then, to use it like this:
my_class a;
a.add<Name>("Peter");
a.add<Foo>(3);
Now, lets dive behind the covers. We are going to use SFINAE combined with enable_if.
template <class Tag, class Type>
class vector
{
template <class T, Return>
struct Enable
{
typedef typename boost::enable_if<
boost::is_same<T,Tag>,
Return
>::type type;
}; // Enable
public:
template <class T>
typename Enable<T,void>::type
add(Type const& i) { m_elements.push_back(i); }
template <class T>
typename Enable<T, Type const&>::type
get(size_t i) const { return m_elements.at(i); }
// You'd better declare a whole lot of other methods if you really want that
// like empty, size and clear at the very least.
// Just use the same construct for the return type.
protected:
vector() : m_elements() {}
vector(vector const& rhs) : m_elements(rhs.m_elements) {}
vector& operator=(vector const& rhs) { m_elements = rhs.m_elements; return *this; }
~vector() {} // Not virtual, because cannot be invoked publicly :)
private:
std::vector<Type> m_elements; // at() is inefficient on lists
};
How does this work ?
Basically when you invoke get<Name> the compiler has 2 alternatives:
vector<Name,std::string>::get
vector<Foo,int>::get
Now, thanks to enable_if, the second alternative is ill-formed (the type deduced cannot be used because Foo != Name), and thus, thanks to SFINAE, this alternative is removed from the list without any complaint.
Then, since there is only one alternative, it gets selected. And of course, since this is done at compile-time, you don't actually have any runtime penalty.
If you want to skip some work (for this type). You could also simply use a simpler construct:
template <class Tag, class Embedded>
class Embed
{
// redeclares same private and protected interface
public:
template <class T>
typename Enable<T,Embedded &>::type get() { return m_element; }
template <class T>
typename Enable<T,Embedded const&>::type get() const { return m_element; }
private:
Embedded m_element;
};
Then you use it like so:
class my_class: public Embed< Names, std::vector<std::string> >,
public Embed<Foo,int>
{
};
my_class a;
std::vector<std::string> const& names = a.get<Names>();
int foo = a.get<Foo>();
a.get<Names>().push_back("Peter");
It's easier since you only provide accessors and don't have to write a whole bunch of methods just to forward the work.
And now that we've work so much, we should ask ourselves: this seems quite practical and generic, surely there is a library or something ?
There is >> Boost.Fusion's map:
class my_class
{
public:
template <class Tag>
typename result_of::at_key<map_type, T>::type &
get() { return at_key<T>(m_data); }
template <class Tag>
typename result_of::at_key<map_type, T>::type const&
get() const { return at_key<T>(m_data); }
private:
// First element of the pair: TAG
// Second element of the pair: Actual type of the data
typedef boost::fusion::map <
std::pair<Name, std::vector<std::string> >,
std::pair<Foo, int>
> map_type;
map_type m_data;
};
The main issue with the macro is: it is solving a problem you don't really have.
It is making a class with list members, where the lists are manipulated directly.
Because we think in OO, members should be encapsulated and we want to use DRY, we come to this construct, while my_class remains really a data-class.
If all the class has to do is contain lists, turn it into a struct and keep the lists public.
This way you have a clean intention and the lists can be accessed the STL-way.
If the class needs to have control over the lists, then you shouldn't expose the lists and the macro's are of little use.
So my code would be (not compiled):
struct my_class {
std::list<int> integers;
std::list<std::string> names;
// ...
};
int main()
{
my_class lists;
lists.integers.push_back(5);
size_t size_names = lists.names.size();
}
Pro:
easily add lists
no code duplication
consistent (STL) interface
straightforward
no macro's
Con:
no data encapsulation, if that would be a requirement
For multiple lists, you can try multiple inheritance with typedefs. Something like:
class my_class : public list_interface<int>, public list_interface<float> {
public:
typedef list_interface<int> Ints;
typedef list_interface<float> Floats;
//...
};
and use like:
my_class a;
a.Ints::add_element(10);
a.Floats::add_element(10.0f);