error C2059

[CPallini]

Quote:

P.S. Of course, 0 values are for testing purpose only.

However they are bad arguments for a (very likely) legitimate macro. If you get the error with such arguments, then we cannot blame the compiler. On the other hand, if you're getting the same error with proper arguments then post such code here.

[mo1492]

This is a combination of all of the above responses. I have compiled this in mfc and it works. struct data { int something; }; struct container { int something_before; struct data data_item; int something_after; }; #define list_entry(ptr, type, member) \ ((type *)((char *)(ptr)-(size_t)(&((type *)0)->member))) // <-- here is the error ... struct data *data_ptr; struct container *cont_ptr = list_entry(data_ptr, struct container, data_item); // This is the code you say is not compiling. #define list_next_entry(pos, member) list_entry(0, 0, 0) // This error is caused by the above line because list_entry(0,0,0) '0' is not a valid data type. const int nTest = list_next_entry(0, 0); // <-- error C2059: syntax error : ')' // If you do this instead... #define list_next_entry(pos, member) list_entry(data_ptr, struct container, data_item) // You will get a new error you will have to resolve in some way. // error C2440: 'initializing' : cannot convert from 'container *' to 'const int' // 1> There is no context in which this conversion is possible // The compiler can't convert a pointer to an int. // Don't know what you are trying to do. const int nTest = list_next_entry(0, 0);

[_Flaviu]

This is legacy code, is not written by me.

[_Flaviu]

With 0 or original argumtents value, the error is the same.

[Lost User]

Then I would suggest you search out every usage of those defines and turn them into real code.

[CPallini]

But I can't help on 0, because the compiler is absolutely right complaining.

[_Flaviu]

I am just trying to integrate a C old code into MFC project. I will try to do what you said, and give you a feedback.

[_Flaviu]

The original code is just like that:

#define list_next_entry(pos, member) list_entry((pos)->member.next, typeof(*(pos)), member)

but the errors is little more then:

error C2100: illegal indirection
error C2059: syntax error : ')'
error C2059: syntax error : 'bad suffix on number'
error C3861: 'typeof': identifier not found

all of them is in the same line ...

[Stefan_Lang]

You can't use -> or * on a value of 0! That causes the compiler errors. Try any non-null value, and at least these compiler errors should go away. Moreover, typeof is not standard C/C++. There are implementations for that in the GCC extension for C, or in the BOOST library. You may need to find the correct BOOST library and include that in your project to make this code work.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

[_Flaviu]

Should I understand that pos has 0 value ? And what I could use instead of

typeof

without using BOOST ?

[_Flaviu]

This is the original code:

#define list_next_entry(pos, member) \
list_entry((pos)->member.next, typeof(*(pos)), member)

[_Flaviu]

So, as a conclusion, the code is:

const int nTest = list\_next\_entry(0, 0);

where list_next_entry is defined:

#define list_next_entry(pos, member) \
list_entry((pos)->member.next, typeof(*(pos)), member)

and list_entry is defined as:

#define list_entry(ptr, type, member) \
((type *)((char *)(ptr)-(size_t)(&((type *)0)->member)))

and for this code I get:

error C2100: illegal indirection
error C2059: syntax error : ')'
error C2059: syntax error : 'bad suffix on number'
error C3861: 'typeof': identifier not found

at line "const int nTest = list_next_entry(0, 0);" I cannot get rid of these errors ... :(

[Lost User]

_Flaviu wrote:

I cannot get rid of these errors

And you will not get rid of them as long as you keep using 0 as the parameters to the list_next_entry call. The define uses specific C/C++ types that are not valid if replaced by 0. In the above case this code translates as follows:

// your statement
const int nTest = list_next_entry(0, 0);

// converts to:
list_entry((0)->0.next, typeof(*(0)), 0)

which makes no sense at all.

[_Flaviu]

Here is the code:

const file\_info\* file = list\_first\_entry(&dir\_list->list, const file\_info, list);
const file\_info\* file2 = td\_list\_next\_entry(file, list);   // <-- errors

and I get the same errors:

error C2059: syntax error : ')'
error C3861: 'typeof': identifier not found

[Victor Nijegorodov]

_Flaviu wrote:

and I get the same errors:

error C2059: syntax error : ')'
error C3861: 'typeof': identifier not found

Of course you must get these error messages! Just because

Quote:

'typeof'

does not exist in MFC (nor in Microsoft C++)

[Victor Nijegorodov]

Did you try to use [decltype (C++) | Microsoft Docs](https://docs.microsoft.com/en-us/cpp/cpp/decltype-cpp?view=vs-2019) rather than typeof?

[_Flaviu]

I have tried:

#define list_next_entry(pos, member) \
list_entry((pos)->member.next, decltype(*(pos)), member)

and I get another errors:

error C2528: 'abstract declarator' : pointer to reference is illegal
error C2528: 'abstract declarator' : pointer to reference is illegal
error C2227: left of '->list' must point to class/struct/union/generic type

seem to go for solving :) ... I guess ...

[Stefan_Lang]

After referring to the SO link you helpfully provided, I finally understand what this macro is intended for. The solution is simple: don't use this macro! It is intended for an extremely specific purpose with specific classes/structs, and it must be used with very specific arguments. If you don't know how to use it correctly, the preprocessor will generate garbagage code, and the compiler unintellegible errors! The purpose is some low level memory address juggling which makes a whole lot of assumptions on behalf of the parameters being passed, without giving the compiler any information about what is going on. This is highly explosive stuff! Obviously you have no idea what parameters to pass and how to use it, probably not even what to use it for. Please don't do that and stop before anyone gets hurt. The only thing that's worse than using #define macros in C++ is using other peoples #define macros without being 200% sure what it does and how it's supposed to be used. If you need to ask what it does or why it doesn't work, then just drop it. Instead, just tell us what goal you intend to achieve, so we can advise you on proper C++ ways of solving your problem.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

[Lost User]

What do you think decltype is going to do for you?

[Stefan_Lang]

Took me a while to figure it out, but decltype(*pos) will give you a reference type, and the parts of the macro using this type to cast to type* then fail, because you can't create a pointer to a reference! However, that's where std::remove_reference comes to the rescue! Try these definitions:

#include // for std::remove_reference

using namespace std;

#define list_entry(ptr, type, member) \
((type *)((char *)(ptr)-(size_t)(&((type *)nullptr)->member)))
#define list_next_entry(pos, member) \
list_entry((pos)->member.next, remove_reference< decltype(*(pos)) >::type, member)

P.S.: are you sure the second #define you posted is correct? I could make up some type definitions that I could use with these macros without encountering compiler errors, but I'm not at all sure the macros would provide a valid pointer for these types! Here's the code I used to validate there are no compiler errors:

struct link
{
struct link* next;
};
struct node
{
int a;
link b;
int c;
};
int main()
{
node y = { 3, nullptr, 2 };
node x = { 5, &y.b, 4 };
node* p = list_next_entry(&x, b);
cout << "value = " << p->a << endl;

return 0;

}

It did compile, but the output was some random number, not 3 as I expected. I suspect that either the structs or that second #define must be defined differently... P.P.S.: it compiles and prints value = 3 as expected.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)