Export template instantiation as C function - c++

I'm creating Haskell bindings for a C library. However, I want to adapt the calling conventions of the library to something more appropriate for Haskell so I've created a template class which has a static method that invokes the correct method when you use it the following way:
Wrap<decltype(&libraryFunction), &libraryFunction>::call(...);
Where ... denotes the arguments. This is great, the compiler has generated the wrapper code for me apparently. I can even take the address of this function using the & operator so it's apparently just a regular function.
However, Haskell cannot use C++ templates so what I'd like to do is to explicitly instantiate this template for the required functions and export them as regular C function which I can reference from Haskell. I know I could just make stubs that manually invoke this static member function but let's say I don't really feel like it.
Any ideas on how to do this? Portable or non-portable, I'd like to know if it's possible.

This cannot be done portably:
[temp]/4
A template, a template explicit specialization, and a class template partial specialization shall not have C linkage.
Note that this also prevents you from taking a pointer to that function letting Haskell call the function via this pointer: It is not guaranteed that e.g. the calling convention of a C++ function (in the C++ program) is the same that Haskell assumes for the function pointer it gets.
What can be done portably is to export a single extern "C" function that translates between C and C++ language linkage. You can pass function pointers or other kinds of identifiers to Haskell, and Haskell calls this single C function passing the identifier of the function it actually wants to call:
std::map<int, void(*)(int32_t)> functions;
extern "C" interface_fct(int id, int32_t arg)
{
functions[id](arg);
}
This of course is not very useful for functions with different parameter sets.
You could write something similar to va_args (but safer) to pass arbitrary arguments, but it might be more useful to write something nonportable.

Related

Are function prototypes necessary for C++?

Basically, I am reading through this book right here and in Section 1.6: Some Differences between C and C++ it is stated:
Another subtle difference between C and C++ is that in a C++ program,
all functions must be prototyped.
I am sure that this is not true from all the C++ programs that I have written. Is this only true for some versions of C++? Is it also true for C?
It has been true of C++ since the beginning (although in C++ it's just called a "declaration", not a "prototype").
As C existed decades ago, it allowed you to call a function without declaring it. You could, however, declare a function if you wanted to--usually to tell the compiler that it had a return type different from what the compiler would deduce on its own. So, in C a function declaration looks something like this:
long f();
Note that empty parens there. That's what separates a function "declaration" from a function "prototype" (though a prototype is basically a superset of a declaration, so a prototype also declares the function in question). A prototype always has something inside the parens to indicate the number and type of parameters the function accepts, on this general order:
short g(int a, double b);
If it doesn't accept any parameters, you have to put in void to indicate that:
int h(void);
If you leave the parens empty, that (as noted above) means it's a function declaration instead of a prototype--and that means you're telling the compiler the function's return type, but you're not telling it anything about the number or type of parameters.
C++ (since before it was called C++, if I recall correctly) has only had one concept instead of the two in C. In C++ every function must be declared--and a declaration always includes the number of parameters, and the type of each. This is absolutely necessary to support (for one obvious example) function overloading, where the correct function to call is determined from the number and types of arguments you pass in the call.
A function definition in C++ also acts as a function declaration. Every function must be declared, but the declaration doesn't have to be separate from the definition.
In reasonably modern C, you normally get pretty much the same--that is, a "new" (i.e., not ancient) type function definition also acts as a prototype for that function. As C was originally defined, it included a syntax for a function definition that looked like this:
int f(a, b, c)
int a;
short b;
long c;
{
// function body here
}
This defines the function, but the compiler treats it only as a function declaration, not a prototype--that is, it tells the compiler the return type, but the number and types of parameters (even though they're specified) are not used by the compiler in the same way they would be with a function prototype. C++ has never used or supported this style of function definition though.

How do to pass variable number of arguments to a function in c++ with no named parameters

I need to write a function which takes a variable number of arguements, its essentially a wrapper around a snprintf like function. I understand how to do this in general as shown in Passing variable number of arguments around.
How ever it appears that, atleast in C, we are required to pass one named parameter to the function before the variable arguments start( Why is void f(...) not allowed in C? ).
However the answers to the same questions say this is easily possible in C++. I was wondering how to achieve this, especially since the method for handling vararg functions seems to be the same as C(va_start, etc). Or are they referring to variadic templates?
So basically is it possible to write a variadic function with no named arguements, as shown below, and if so, how?
void f(...){
...
}
p.s. I would like to do this with c++03 if possible.
While C++ language allows writing functions whose parameter list consist only of ..., the language provides no means for accessing the arguments in functions declared with (...) parameter list.
Basically such functions in C++ exist for their behavior in overload resolution (... parameters match any parameter type). The body of such function will not have access to the arguments. If you see no uses for such functions, just don't use them.
Such functions are often used in non-executed context in some well-known template meta-programming techniques, in which case they don't even have to be defined - a mere declaration is sufficient.

Usage of variadic functions

What are the benefits of using
variadic functions
void fun(int i, ...);
instead of passing a pointer to an array?
void fun(int i*);
When are variadic functions preferred?
You have to explicitly create the array first. In addition, it would be problematic if you wanted to specify parameters of different types.
Variadic functions do not require creating any array and they can deal with different types.
As an example, I could not bare to use the printf-functions if I always had to create some array first.
On the other hand, in most cases it's just a form of syntactic sugar, I suppose.
Pointer to array assumes predefined type of the parameter (or struct, if its several different types).
Variadic functions are used when you don't know ahead of time what would the type of the parameter be, and you use a hint of the predefined parameters to get that knowledge (like the format string for printf).
Also, you don't want to pass an array in a variadic function, as you would also want to pass in its size. e.g:
myfunction(int *pArray, int NumElements)
The main point, though, is that variadic functions allow many different types to be passed.
I would suggest you to simply not use variadic functions. However they can be useful. For example in template metaprogramming techniques to implement compile-time querys.
As the compiler can't verify that a given call to a variadic
function passes an appropriate number of arguments or that those
arguments have appropriate types. Consequently, a runtime call to a
variadic function that passes inappropriate arguments yields
undefined behavior.
In pure C environments variadic functions have no alternative but in
c++ you are able to find object oriented alternatives (also one
issue).
Variadic functions are not trivial to implement. Failing to initialize ap, calling va_arg() one time to many or omitting the va_end(ap) call can crash your program.
Example of being close to undefined behavior:
It was my job to crate a object-oriented wrapper around the sqlite C api.
Here I am: I created a fancy interface for executing sqlite querys it was something like that:
void Query(const std::string& Query, void* Arguments, ...);
This is on one side awesome and fancy but: You immediately encounter undefined behavior if your parameter types are wrong. For example:
db->Query("select * from northwind where id = ?", "Peter"); // espects int as type
Will result in undefined behavior.

TI DSP: interfacing C++ and assembly

I posted this Q to TI's 28xx DSP forum but haven't heard a response and figured maybe someone here might know.
I know how to write functions in assembly so that they are C-callable; if the C-callable name is foo() then the assembly function is named _foo().
What if I want to use C++ and optimize a class method in assembly? How do I do that? I assume the only major issues are:
naming
accessing the "this" pointer
accessing class members by somehow knowing offsets
and if I don't want to worry about the last two, then perhaps I would write a static member function and do this:
class MyClass
{
int x;
static int _doSomething(int u); // implement this in assembly
public:
inline void doSomething() { x = _doSomething(x); }
// lightweight C++ wrapper to handle the class member / "this" pointer stuff
};
The this pointer gets passed as an additional argument to the function, using the standard calling convention on your platform. On all the platforms I'm familiar with it is passed as the first argument, but I don't do a lot of C++ coding, so I'm not sure if this is guaranteed by the standard. You can always disassemble some C++ code on your platform to confirm.
The C++ symbol naming is rather more painful than in C, and varies from compiler to compiler. I suppose you could figure out the right symbol name to use by disassembling a compiled function definition, just make sure that: the function is a member of the right class, and has the right number and type of arguments.
Unless you really need to reproduce a C++ function in situ, I would probably just make a standard C function and do the usual extern "C" { ... } around its declaration.
Does your compiler have an inline assembly syntax? If you have that, it may be the easiest option, and you can let the compiler handle the function naming and call syntax pieces.
Alternately, Stephen's suggestion of writing the C++ method as an inlined wrapper around a "simple" C function call is a good one. (You probably want to make it just a plain function, not a static member function as in your post, to get a simple C interface to it.)
I would find the compiler-dependent flag and write the assembly within the C++ function. Usually there are ways to reference local variables from within the assembly section.

Is there any way to programmatically determine in C/C++ how many parameters a Lua function expects?

Is there a way to determine how many parameters a Lua function takes just before calling it from C/C++ code?
I looked at lua_Debug and lua_getinfo but they don't appear to provide what I need.
It may seem a bit like I am going against the spirit of Lua but I really want to bullet proof the interface that I have between Lua and C++. When a C++ function is called from Lua code the interface verifies that Lua has supplied the correct number of arguments and the type of each argument is correct. If a problem is found with the arguments a lua_error is issued.
I'd like to have similar error checking the other way around. When C++ calls a Lua function it should at least check that the Lua function doesn't declare more parameters than are necessary.
What you're asking for isn't possible in Lua.
You can define a Lua function with a set of arguments like this:
function f(a, b, c)
body
end
However, Lua imposes no restrictions on the number of arguments you pass to this function.
This is valid:
f(1,2,3,4,5)
The extra parameters are ignored.
This is also valid:
f(1)
The remaining arguments are assigned 'nil'.
Finally, you can defined a function that takes a variable number of arguments:
function f(a, ...)
At which point you can pass any number of arguments to the function.
See section 2.5.9 of the Lua reference manual.
The best you can do here is to add checks to your Lua functions to verify you receive the arguments you expect.
You can determine the number of parameters, upvalues and whether the function accepts variable number of arguments in Lua 5.2, by using the 'u' type to fill nups, nparams, isvararg fields by get_info(). This feature is not available in Lua 5.1.
I wouldn't do this on the Lua side unless you're in full control of Lua code you're validating. It is rather common for Lua functions to ignore extra arguments simply by omitting them.
One example is when we do not want to implement some methods, and use a stub function:
function do_nothing() end
full_api = {}
function full_api:callback(a1, a2) print(a1, a2) end
lazy_impl = {}
lazy_impl.callback = do_nothing
This allows to save typing (and a bit of performance) by reusing available functions.
If you still want to do function argument validation, you have to statically analyze the code. One tool to do this is Metalua.
No, not within standard Lua. And is Aaron Saarela is saying, it is somewhat outside the spirit of Lua as I understand it. The Lua way would be to make sure that the function itself treats nil as a sensible default (or converts it to a sensible default with something like name = name or "Bruce" before its first use) or if there is no sensible default the function should either throw an error or return a failure (if not name then error"Name required" end is a common idiom for the former, and if not name then return nil, "name required" end is a common idiom for the latter). By making the Lua side responsible for its own argument checks, you get that benefit regardless of whether the function is called from Lua or C.
That said, it is possible that your modules could maintain an attribute table indexed by function that contains the info you need to know. It would require maintenance, of course. It is also possible that MetaLua could be used to add some syntax sugar to create the table directly from function declarations at compile time. Before calling the Lua function, you would use it directly to look up any available attributes and use them to validate the call.
If you are concerned about bullet-proofing, you might want to control the function environment to use some care with what (if any) globals are available to the Lua side, and use lua_pcall() rather than lua_call() so that you catch any thrown errors.
The information you ask for is not available in all cases. For example, a Lua function might actually be implemented in C as a lua_CFunction. From Lua code there is no way to distinguish a pure Lua function from a lua_CFunction. And in the case of a lua_CFunction, the number of parameters is not exposed at all, since it's entirely dependent on the way the function is implemented.
On the other hand, what you can do is provide a system for functions writers (be it in pure Lua or in C) to advertise how many parameters their functions expect. After creating the function (function f(a, b, c) end) they would simply pass it to a global function (register(f, 3)). You would then be able to retrieve that information from your C++ code, and if the function didn't advertise its parameters then fallback to what you have now. With such a system you could even advertise the type expected by the parameters.