Do different files have different scopes? - c++

In file1.cc I write
int i = 0;
while in file2.cc I write
#include <iostream>
int i = 1;
int main()
{
std::cout<< i << std::endl;
return 0;
}
In MacOS the compiler reports
duplicate symbol _i in:
/var/folders/wn/q9648wb507j9l504vp2d_dwm0000gn/T/file1-bb8eca.o
/var/folders/wn/q9648wb507j9l504vp2d_dwm0000gn/T/file2-b5e667.o
ld: 1 duplicate symbol for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
But isn't it that different files have their different scopes, so that we can define a global variable in file2 with the same name as in file1?
Moreover, if different files are in the same scope, then why is it illegal to transform file2.cc as:
#include <iostream>
int main()
{
std::cout<< i <<std::endl;
return 0;
}

Multiple definitions of same global variable are not allowed. You must mark your variables static, const, or place them in an anonymous namespace. Else they are in the global scope.
And your second question: The example code does not compile, because the compiler does not see the global variable from the other file, because a compiler compiles one translation unit (C++ file) at a time. This is why you need to add extern i;. This tells the compiler, that the variable will be defined in another translation unit.
After all files have been compiled successfully, they will be linked. Linker checks all translation units. When you get an error about multiple definitions, this error occurs during linking.

The global scope for files linked in the same executable is the same. To avoid that you could declared it static, but that is deprecated in c++.
The correct way to separate the scope of the two files is to use namespaces.

Your variable is in the same scope. You have to use namespaces in order to declare to variables with the same name.

Related

Multiple includes of same header file in project: C vs C++

Here I have an example project with two source files and a header file, as follows:
main.c:
#include<stdio.h>
#include "personal.h"
int main(){
i = 5;
printf("Value is %d\n",i);
return 0;
}
sub.c:
#include "personal.h"
// do nothing
and finally personal.h:
#pragma once
int i;
Each of the .c file includes the personal.h, which is `guarded'. I compile with gcc, all goes fine:
>gcc sub.c main.c -o out
>./out
Value is 5
But with g++, this happens:
>g++ sub.c main.c -o out
/tmp/cctYwVnO.o:(.bss+0x0): multiple definition of `i'
/tmp/ccPElZ27.o:(.bss+0x0): first defined here
collect2: error: ld returned 1 exit status
Is there anything fundamentally different between C++ and C in terms of how files are linked, preprocessor activity etc? I tried the same with other compilers like clang and the same happens. I am perhaps missing something silly here.
In C,
int i;
is a tentative definition. By the virtue of inclusion, you have a tentative definition for i in two compilation units. The C standard allows to have tentative definitions in multiple compilation units, but does not mandate that implementation accept that. The custom behavior for Unix C compilers is to allow it but gcc has an option (-fno-common) to prevent it and generate an error at link time (so that you can check the code for compilers, as I think Microsoft one, which does not allow it or for platforms for which that allow better code -- I know of none but that's a rationale given by GCC documentation).
IIRC, C++ has no such allowance.
Note that you probably want a declaration and not a definition in headers. Thus for the i above it should be
extern int i;
in the header and one
int i;
at global scope in one .c file.
sub.c will include personal.h and will create the variable i in global scope. Similarly, main.c will also include personal.h and create variable i in global scope. Eventually, when you link, there are two definitions of i in the global scope and hence the error.

g++ fails to link .o files into an executable

I am doing an example drill in the textbook I am using to learn from. All I need to do is compile, link and run the following 3 files:
//file my.h
extern int foo;
void print_foo();
void print(int);
my.h is a simple header file that declares the two functions and a 'global' int foo, with no initial value.
//file my.cpp
#include "my.h"
#include "std_lib_facilities.h" //not included but not source of error
void print_foo()
{
cout << foo << endl;
}
void print(int i)
{
cout << i << endl;
}
my.cpp contains the implementation of the functions included from my.h. std_lib_facilities.h is a file from the textbook, and is not the source of error (according to g++). I can edit it into the body of the question if needed.
//file use.cpp
#include "my.h"
#include <iostream>
int main() {
foo = 7;
print_foo();
print(99)
char cc; cin >> cc;
return 0;
}
use.cpp serves as the main implementation file in this program, and tries to use all three declared & defined objects.
I took the two step command approach to build using g++. First, I compiled both .cpp files:
g++ -c my.cpp use.cpp
which created two object files, my.o and use.o. I used the following command to link them:
g++ -o myprog my.o use.o
giving me this error:
Undefined symbols for architecture x86_64:
"_foo", referenced from:
print_foo() in my.o
_main in use.o
(maybe you meant: __Z9print_foov)
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
I have tried putting
int foo;
into my.h instead of
extern int foo;
which gave me the same error.
I have tried using the
-std=c++11
flag as well which resulted in the same error.
I am using a MacBook Pro with the latest macOS (just updated in fact), if that helps with interpreting the error message.
I have tried to initialize foo, which didn't change anything.
In addition, I have tried updating the command line tools, same error.
From what I understand, the error is telling me that, even though my.h is included in both files, neither one can actually implement any function using the foo variable (which it calls _foo), despite it being explicitly declared in my.h. My guess is that the linker is using the wrong names under the hood, which make it impossible to link into an executable. This comes from the fact that the error mentioned a
__Z9print_foov
which exists nowhere in any of the files.
It almost seems like a g++ or macOS/Command Line Tools bug at this point. I don't want to add the declarations each time, because that creates duplicate symbol errors anyway. Putting my.cpp and use.cpp into one file would probably link properly, but I need to make sure that I can actually link multiple cpp files, because I will eventually (hopefully) be working with multiple cpp files that need to be linked. Any help is appreciated!
Here you declare a variable:
extern int foo;
and you use the variable:
cout << foo << endl;
but you did not define the variable anywhere. The linker error says that the linker could not find the variable's definition. To fix this, put int foo; at file scope in one of the .cpp files.
In the question you say that changing extern int foo; to int foo; gives the same error. However if you look more carefully at the error message I think you will find that it gives a different one, about multiple definitions.
I suggest to compile in two commands g++ -Wall -c my.cpp (that gives a my.o) and g++ -Wall -c use.cpp (giving use.o), then link a program with g++ my.o use.o -o myprog. Actually you should write a Makefile (see this for inspiration) and simply run make
Your translation units my.cpp and use.cpp are both declaring some extern int foo; variable which is never defined. So you need to define it in one single file (but not in others!), probably by adding (into my.cpp alone for example)
int foo;
(without the extern) or even with some explicit initial value e.g. int foo = 34;
This comes from the fact that the error mentioned a __Z9print_foov which exists nowhere
It is a mangled name, which is referenced (but not defined) in both object files (see also this).
It almost seems like a g++ or macOS/Command Line Tools bug at this point
You are very unlikely to find bugs in compiler tools (both GCC & Clang/LLVM are extremely well tested; since they are multi-million lines free software, they do have residual bugs, but you have more chances to win at the lottery than to be affected by a compiler bug). I'm coding since 1974, and it happened to me only once in my lifetime. A more realistic attitude is to be more humble, and question your own code (and knowledge) before suspecting the compiler or build chain.
BTW, always compile first with all warnings and debug info (e.g. g++ -Wall -g and perhaps also -Wextra). Use the gdb debugger. When you are convinced that your code has no bugs, you might benchmark it by asking the compiler to optimize (so use g++ -Wall -O2 perhaps also with -g to compile).
Read also the linker wikipage. Dive into your C++ textbook (see also this site and the C++11 standard, e.g. n3337 draft) to understand the difference between declaring and defining some variable or function. You generally declare a global extern variable in some common header (included in several translation units), and define it once somewhere else, but the good practice is to avoid having lots of global variables. See also C++17 new inline variables.

Undefined reference error when using a simple class inside my function

I am getting nuts with this error so I thought some of more experienced developers can help me in this regard.
I am trying to compile a sample project which uses a C++ library (named Poco). My project is linked to compiled poco libraries.
Below is my (most simplified) code:
#include "Poco/UUID.h"
class x
{
void func1()
{
new Poco::UUID(); //A
}
};
void func1()
{
new Poco::UUID(); //B
}
Now when above code is compiled, line 'A' has no error but for line 'B' linker says:
undefined reference to `Poco::UUID::UUID()'
What is the reason? When I instantiate a class from external lib in a class method no error occurs but the same code in a function produces linker error? (When I comment line B, no error occurs and linker output files are generated)
My configuration: Win7/g++/CodeLite/MinGW-4.7.1
*Update 2:*Thanks. My problem is now resolved and the issue is that I had compiled library using MSVC compiler while my application was being compiled using g++ (both under Windows platform). So I re-compiled library using g++ and everything works fine now.
Update 1: here is my IDE's output when I build my project:
C:\Windows\system32\cmd.exe /c "mingw32-make.exe -j 4 -e -f "dll1.mk" all"
----------Building project:[ dll1 - Debug ]----------
g++ -shared -fPIC -o ./Debug/dll1.so #"dll1.txt" -L. -Lc:/poco/lib -lPocoFoundationd
./Debug/PluginLibrary.o: In function `Z5func1v':
C:/Users/PARS/Documents/codelite/workspace1/dll1/PluginLibrary.cpp:12: undefined reference to `Poco::UUID::UUID()'
collect2.exe: error: ld returned 1 exit status
mingw32-make.exe: *** [Debug/dll1.so] Error 1
dll1.mk:77: recipe for target `Debug/dll1.so' failed
1 errors, 0 warnings
Your member function x::func1() is never ODR-used in that compilation unit (source file). Most compilers only generate compiled code for a member function defined inside the class definition if that member function is ODR-used within the compilation unit that is being compiled. Suppose some other source file does use x::func1(). If you compile that other source file, the compiler will produce object code for x::func1() in the object file that corresponds to that other source file.
The compiler can get away with bypassing the process of generating compiled code for x::func1() here because the class definition has to be the same across all compilation units. If you compile some other source file that has a different definition of class x you have violated the one definition rule. This is undefined behavior and no diagnosis is required.
If no source file uses x::func1() you have some dead code that just never happens to be compiled. The code has an error but it's never detected.
The compiler cannot get away with bypassing generating compiled code for the free function func1(). This function has external linkage; there's no way the compiler can tell if it might be used somewhere else. The compiler must generate compiled code for that free function.
Here's a minimum working example:
class Missing {
public:
Missing();
int value;
};
class UsesMissing {
public:
int use_missing () {
Missing missing;
return missing.value;
}
int dont_use_missing () {
return 0;
}
};
#ifdef DEFINE_USE_MISSING
int use_missing () {
Missing missing;
return missing.value;
}
#endif
int main () {
UsesMissing test;
#ifdef USE_MISSING
return test.use_missing();
#else
return test.dont_use_missing();
#endif
}
Compile with neither DEFINE_USE_MISSING or USE_MISSING defined and this compiles and links just fine with g++ and clang++. Define either one of those flags and the file fails in the link step because of the undefined reference Missing::Missing().
You should link with the correct library to fix your link (see Poco docu for the correct one).
func1 has extern linkage and so linker need Poco::UUID
whereas X::func1 is inline/private/unused.
if you use static foo1() or inline foo1() the linker error disappears
if you use x::func1 or implement x::func1 outside of the class x{}; the error linker appears

Linking with a library from external directory

Edit: I have updated my question with changes I've made, upon answers.
I'm trying to link to a little library that I've wrote to learn ho this is done with C++ with no luck. G++ is complaining with undefined reference.
The root directory of library I want to link is in directory ~/code/gklib/cxx/. The structure of this directory is as follows:
~/code/gklib/cxx/
|
|`-> gk.{hh,cc}
|`-> libgk.o
|
`-> lib/
|
`-> libgk.a
I have compiled gk.cc with -c flag, then transformed the resulting object file to a .a file with ar rvsc lib/libgk.a libgk.o.
The client to this library is at folder ~/code/cpp. In this directory I compiled some_stuff.cc to an object file again, then I tried to link to it with this command:
$ cxx some_stuff.o -L../gklib/cxx/lib -lgk -o some_stuff
I get this error:
some_stuff.o: In function `main':
some_stuff.cc:(.text+0x49): undefined reference to `void GK::algorithms::insertionSort<int, 5ul>(int*)'
collect2: error: ld returned 1 exit status
These are contents of these files:
~/code/cpp/some_stuff.cc
#include <cstdlib>
#include <iostream>
#include <gk.hh>
using namespace std;
int main(int argc, char **argv) {
int i = -1;
int arr[5] = { 3, 4, 2, 1, 5 };
const size_t len = sizeof(arr)/sizeof(int);
GK::algorithms::insertionSort<int, len>(arr);
while(++i < 5)
cout << arr[i] << " ";
cout << endl;
exit(EXIT_SUCCESS);
}
~/code/gklib/cxx/gk.cc
#include "gk.hh"
template<class T, size_t len>
void GK::algorithms::insertionSort(T arr[len]){
// insertion sort
}
~/code/gklib/cxx/gk.hh
#pragma once
#include <cstdlib>
#define NAMESPACE(ns) namespace ns {
#define END_NAMESPACE(ns) }
NAMESPACE(GK)
NAMESPACE(algorithms)
template<class T, size_t len>
extern void insertionSort(T arr[len]);
END_NAMESPACE(algorithms)
END_NAMESPACE(GK)
I've tried many variations on my commands with no result. Internet is full of tutorials and forums with instructions those did not work for me. This code ran perfectly when all the stuff was in one file. How can I resolve this problem? Thanks in advance.
I think it's more something like:
cxx some_stuff.o -L$HOME/gklib/cxx/lib -B../gklib/cxx/lib -lgklib -o some_stuff
-lgklib, not -Igklib (-I option specify an include folder)
but you'll have to rename your gklib.a by libgklib.a
Maybe you can even remove -B../gklib/cxx/lib, just try it out :)
I see several problems: in order:
If this is your exact command line, the -L option doesn't
point to the structure you've shown above (where there is no
cxx in the path).
I don't know why you're using -B. This tells the compiler
where to look for its libraries, etc. Normally, this is only
necessary when you want to test parts of the compiler you've
modified.
I don't see where you've specified to link against the
library. Since the library doesn't respect the usual naming
conventions (libname.a), you'll have to
specify it directly (in the same way you'd specify an object
file), and the -L option isn't used. Alternatively, you name
it libgk.a, or something like that, and add a -lgk to the
command line, after your object files.
Finally, the error messages refer to an instantiation of
a template. This typically occurs because the implementation of
the template is in a source file, not in the header. The
standard requires that the implementation be visible when it
triggers the instantiation of a template. The way g++ (and
almost every other compiler) works is that if the implementation
isn't visible, they suppose that the template will be
instantiated in another translation unit, and output external
references to it. But if the implementation is in a source,
there will be no instantiation, and you'll get a linker error.
(You'll also not find the assembler for the instantiation
anywhere, either, because the template hasn't been
instantiated.) You should include the source code at the bottom
of your header, and not compile it separately.
I have solved this problem, with the help of this and this questions. The problem lies in the fact that void insertionSort(T *) is a template function, and template functions can only be implemented in header files. The reason for this is, the compiler needs to reach the definition of the to create a new function for each call to it with a different type as the argument to template. I have moved the definition of this function to gk.h, which resulted in a successfull compilation and linkage.

What is the best way to eliminate MS Visual C++ Linker warning : "warning LNK4221"?

I have a CPP source file that uses #if / #endif to compile out completely in certain builds. However, this generates the following warning.
warning LNK4221: no public symbols found; archive member will be inaccessible
I was thinking about creating a macro to generate a dummy variable or function that wouldn't actually be used so this error would go away but I want to make sure that it doesn't cause problems such as using the macro in multiple files causing the linker to bomb on multiply defined symbols.
What is the best way to get rid of this warning (without simply suppressing the warning on the linker command line) ?
FWIW, I would be interested in knowing how to do it by suppressing the warning on the linker command line as well but all my attempts there appear to be simply ignored by the linker and still generate the error.
One other requirement: The fix must be able to stand up to individual file builds or unity build (combine CPP file builds) since one of our build configurations is a bulk build (like a unity build but groups of bulk files rather than a single master unity file).
Use an anonymous namespace:
namespace { char dummy; };
Symbols within such namespace have external linkage, so there will be something in the export table. On the other hand, the namespace name itself will be distinct (you can think of it as "randomly generated") for every translation unit, so no clashes.
OK, the fix I am going to use is Pavel's suggestion with a minor tweak. The reason I’m using this fix is it’s an easy macro to drop in and it will work in bulk-builds / unity-builds as well as normal builds:
Shared Header:
// The following macro "NoEmptyFile()" can be put into a file
// in order suppress the MS Visual C++ Linker warning 4221
//
// warning LNK4221: no public symbols found; archive member will be inaccessible
//
// This warning occurs on PC and XBOX when a file compiles out completely
// has no externally visible symbols which may be dependant on configuration
// #defines and options.
#define NoEmptyFile() namespace { char NoEmptyFileDummy##__LINE__; }
File that may compile out completely:
NoEmptyFile()
#if DEBUG_OPTION
// code
#endif // DEBUG_OPTION
(Though the discussion is already old and I cannot comment directly #Adisak's answer), I guess some additional macro expansion magic is needed for this to work:
#define TOKENPASTE(x, y) x ## y
#define TOKENPASTE2(x, y) TOKENPASTE(x, y)
#define NONEMPTY_TRANSLATION_UNIT char TOKENPASTE2(NoEmptyFileDummy, __LINE__);