Multiple definitions error for all function definitions in header

Ok, this is weird...
My compiler is producing a "multiple definition" error for every single function definition inside a header file that I wrote. Each function definition is a member function of a class which is defined in the same file. I put the function definitions after the class definition.

My header has include guards (i.e. #ifndef, #define, and #endif) wrapped around the whole code, and needless to say, I don't have any duplicated function definitions.

Anyone have any idea why its doing this??

Thanks in advance.

EDIT: Another interesting thing to note is, when I click on one of the multiple definition errors, it brings up the 'new' file, with the cursor on the line which says:

Code:

---

inline void* operator new(std::size_t, void* __p) throw() { return __p; }

---

What does this have to do with the "new" operator?

Separate your header file from that of defintion file and the main file..

Quote:

---

Originally Posted by Mr.777

Separate your header file from that of defintion file and the main file..

---

Is that really necessary? As this is a generated file we're discussing, I want to keep the code in a single file, if possible.

Isn't there someway to make it work with the function definitions in the header file, along with the class definition?
A google search seems to suggest using the 'inline' keyword.

Yeah inline functions are those that are decided by the compiler, not by the programmer...
And these functions have just one or two lines of code..

So, now it depends what kinda functions you want to code....
If you want your functions to limit their functionality for one or two lines, try using inline functions...
Hope it will help.

Quote:

---

Originally Posted by Mr.777

Yeah inline functions are those that are decided by the compiler, not by the programmer...
And these functions have just one or two lines of code..

So, now it depends what kinda functions you want to code....
If you want your functions to limit their functionality for one or two lines, try using inline functions...
Hope it will help.

---

No can do...
My functions have many lines of code, and there's no way around that.

EDIT: A quick question about 'inline' though...does it matter if you put the keyword before the function declaration or before the definition (or is both allowed)?

I think both are allowed, but putting the keyword 'inline' doesn't guarantee you that now it's an inline function...
By the way, why you don't want to separate files??
May be there are many other options to keep it in a single file..

Quote:

---

Originally Posted by Mr.777

I think both are allowed, but putting the keyword 'inline' doesn't guarantee you that now it's an inline function...

---

Yeah, I know. I skimmed through a page on 'inline' which came up in a Google search, and learned that the compiler makes decisions about what functions to make inline, and what functions to not make inline, which I guess is implementation-specific.
I went ahead, and added the inline keyword before all function declarations in my generated file, and that got rid of the "multiple definition" errors.
Now its returning a "collect2: ld returned 1 exit status" instead. :(

And I actually mis-phrased my above post. I don't actually have many lines in all functions of my generated file. I only have a bunch of lines for the constructor and two other functions. All the other ones only have one line of code inside the braces (not counting the empty line before and after).

Quote:

---

By the way, why you don't want to separate files??
May be there are many other options to keep it in a single file..

---

Well, for one, it would require serious modification (ok, maybe I'm exaggerating just a little...) to the code for the program that generates the file, and since I already made way too many changes after calling it 'done' the first ten-thousand times, I would like to think for once this program is now actually done (except for minor updates here and there, and if I want to add a feature). :p For two, I don't know, one generated file is easier to keep track of than two.

Lol..... I think, you should try learning, handling many files with your single project.

So, what's the problem now?

If your functions got only one or two lines, make them inline and can you work without writing your own constructor because compiler can call default constructor in object request of that class...

Quote:

---

Originally Posted by Mr.777

Lol..... I think, you should try learning, handling many files with your single project.

---

Oh, I will...just not in THIS program. :p

Quote:

---

So, what's the problem now?

---

Quote:

---

Originally Posted by Programmer_P

Now its returning a "collect2: ld returned 1 exit status" instead. :(

---

Quote:

---

If your functions got only one or two lines, make them inline and can you work without writing your own constructor because compiler can call default constructor in object request of that class...

---

No, my code wont work without writing my own constructor, because there are many important class variables which are assigned their values inside the constructor.

Hmmm.... I have no idea of writing function definition inside header file... May be any solution exists but i got no such idea...
Anyways, for good practice i will recommed you to make separate files...

Quote:

---

Originally Posted by Programmer_P

I went ahead, and added the inline keyword before all function declarations in my generated file, and that got rid of the "multiple definition" errors.
Now its returning a "collect2: ld returned 1 exit status" instead. :(

---

That usually means you have an unresolved symbol (e.g. the linker is looking for a function, or maybe a static variable, that is not defined anywhere).

Quote:

---

Originally Posted by Programmer_P

And I actually mis-phrased my above post. I don't actually have many lines in all functions of my generated file. I only have a bunch of lines for the constructor and two other functions. All the other ones only have one line of code inside the braces (not counting the empty line before and after).

---

Yeah, OK. Your problem is not actually related to size of functions - the size of the function is only one factor that may cause the implementation not to inline a function.

Your problem is probably related to a definition being seen in multiple object files by the linker. From what you are describing, you are #include'ing your header from multiple source files (eg .cpp files), which means that any definitions in that header files appear in multiple object files.

Declaring your functions inline should address that problem. While it is true that the implementation (compiler, linker, etc as a suite) might decide not to inline a function that your code declares as inline ("inline" is a hint, not a directive) then the implementation is required to sort things out.

Without seeing samples of actual code (a sample header, and a sample of two source files that include that header, and exhibit the problem) it is harder to give more specific advice.

Quote:

---

Right 98% of the time, and don't care about the other 3%.

---

No offense but shouldn't it be 2%??
lol.... just a blunt humour.... hope you'll not mind that...

Quote:

---

Originally Posted by Mr.777

Yeah inline functions are those that are decided by the compiler, not by the programmer...
And these functions have just one or two lines of code..

So, now it depends what kinda functions you want to code....
If you want your functions to limit their functionality for one or two lines, try using inline functions...
Hope it will help.

---

There is no limit to the size of lines of code in a inline function.
Although having lots of lines of code in a function makes it less likely the compiler will make it inline.

Quote:

---

Originally Posted by grumpy

That usually means you have an unresolved symbol (e.g. the linker is looking for a function, or maybe a static variable, that is not defined anywhere).

---

Well, I am 100% sure that all functions I declared are defined somewhere. And I don't have any static variables in my class.

Quote:

---

Yeah, OK. Your problem is not actually related to size of functions - the size of the function is only one factor that may cause the implementation not to inline a function.

Your problem is probably related to a definition being seen in multiple object files by the linker. From what you are describing, you are #include'ing your header from multiple source files (eg .cpp files), which means that any definitions in that header files appear in multiple object files.

---

Actually, no I'm not. I only include the header in another header, which is then included in a single source file (and that only once). Also, I have header guards in the generated header file that we're discussing, so I don't see why that would be a problem even if I had included it multiple times.

Quote:

---

Declaring your functions inline should address that problem. While it is true that the implementation (compiler, linker, etc as a suite) might decide not to inline a function that your code declares as inline ("inline" is a hint, not a directive) then the implementation is required to sort things out.

Without seeing samples of actual code (a sample header, and a sample of two source files that include that header, and exhibit the problem) it is harder to give more specific advice.

---

Well, I attached the header with the function definitions in it, so you can take a look.

Header guards guard against including the same file multiple times in the same source file, not against including it into multiple source files (once).
The single header is not enough to identify the problem. You should provide the smallest possible compilable example that demonstrates the problem.

Quote:

---

Originally Posted by Elysia

Header guards guard against including the same file multiple times in the same source file, not against including it into multiple source files (once).

---

Right. But I only included it in one source file, not multiple source files.

See edit.

Quote:

---

Originally Posted by Programmer_P

Well, I am 100% sure that all functions I declared are defined somewhere. And I don't have any static variables in my class.

---

When there is a dispute between claims of a programmer and the results of the build process, I would be willing to bet the programmer is mistaken.

Quote:

---

Originally Posted by Programmer_P

Actually, no I'm not. I only include the header in another header, which is then included in a single source file (and that only once). Also, I have header guards in the generated header file that we're discussing, so I don't see why that would be a problem even if I had included it multiple times.

---

So you tell us. However, you're the one withholding information you deem irrelevant, and forcing us to guess what relevant information you've withheld.

You don't need to provide your complete project, but a SMALL and complete example of code that illustrates your concern will be useful. In the process of creating such an example (cutting out the irrelevant bits and testing) you might even find the actual problem yourself.

Providing a header file is not enough. You are reporting problems in the build process, and a header file can't usually be compiled in isolation. The problem you have is almost certainly due to some interaction between your header file and the source file (or files, although I note your repeated claim there is only one) that #include it.

Your header file is at odds with quite a few good coding practices.

1) NEVER employ "using namespace std;" in a header file. That is an effective way to break code that #include's that header file - often in a manner that is unsolvable. Yes, that means you need to sprinkle "std::" prefixes everywhere.

2) You have declared member functions of your class as inline. However, it is the definitions (i.e. implementations) of the functions that need to be inline.

3) It is rather pointless (and also expensive) to throw an exception and then immediately catch it in order to rethrow it.

4) Your class effectively uses member functions to return references to private members. That is essentially bypassing the benefits of having private members.

Quote:

---

Originally Posted by grumpy

2) You have declared member functions of your class as inline. However, it is the definitions (i.e. implementations) of the functions that need to be inline.

---

As far as I am aware, you only need to declare a function inline for it to be inline. You don't need the inline keyword in the definition or implementation.
Thus, Programmer_P's way of defining the functions inline in the class is valid, AFAIK.
(It being a good idea is another matter.)

Quote:

---

Originally Posted by Elysia

(It being a good idea is another matter.)

---

That was actually my point.

Having little mismatches between declaration and definition is a good way to introduce little mistakes. Such as declaring one thing, but the subsequent definition defines something else.

Quote:

---

Originally Posted by grumpy

When there is a dispute between claims of a programmer and the results of the build process, I would be willing to bet the programmer is mistaken.

So you tell us. However, you're the one withholding information you deem irrelevant, and forcing us to guess what relevant information you've withheld.

You don't need to provide your complete project, but a SMALL and complete example of code that illustrates your concern will be useful. In the process of creating such an example (cutting out the irrelevant bits and testing) you might even find the actual problem yourself.

Providing a header file is not enough. You are reporting problems in the build process, and a header file can't usually be compiled in isolation. The problem you have is almost certainly due to some interaction between your header file and the source file (or files, although I note your repeated claim there is only one) that #include it.

Your header file is at odds with quite a few good coding practices.

1) NEVER employ "using namespace std;" in a header file. That is an effective way to break code that #include's that header file - often in a manner that is unsolvable. Yes, that means you need to sprinkle "std::" prefixes everywhere.

2) You have declared member functions of your class as inline. However, it is the definitions (i.e. implementations) of the functions that need to be inline.

3) It is rather pointless (and also expensive) to throw an exception and then immediately catch it in order to rethrow it.

4) Your class effectively uses member functions to return references to private members. That is essentially bypassing the benefits of having private members.

---

1) Oops...forgot I did that. I'll fix that.
2) I don't think that's true. It can be written before either or both. But I went ahead and put it before both shortly after attaching the file here.
3) While that may be, I did that only to get rid of a compiler error saying it expected a catch before the next coding statement. It seems that that is a requirement.
4) No, my functions all return const references.

Quote:

---

Originally Posted by Elysia

As far as I am aware, you only need to declare a function inline for it to be inline. You don't need the inline keyword in the definition or implementation.
Thus, Programmer_P's way of defining the functions inline in the class is valid, AFAIK.
(It being a good idea is another matter.)

---

Quote:

---

Originally Posted by grumpy

That was actually my point.

Having little mismatches between declaration and definition is a good way to introduce little mistakes. Such as declaring one thing, but the subsequent definition defines something else.

---

As mentioned above, I changed it already to have inline before both declarations and definitions.

And just to clear up another point, I am not ignoring the request for a small compilable example. Its just I'm a little bit busy right now, and it'll have to be sometime later.

Quote:

---

Originally Posted by Programmer_P

3) While that may be, I did that only to get rid of a compiler error saying it expected a catch before the next coding statement. It seems that that is a requirement.

---

That should not be a requirement. Possibly you did not understand the compiler error or you did something weird?
You are never required to catch an exception.

Possibly you also tried to add a try block. Each try block requires at least one catch block. However, you do not need a try block to throw exceptions.
If you are not going to catch an exception, then don't add a try block.

Quote:

---

Originally Posted by Elysia

That should not be a requirement. Possibly you did not understand the compiler error or you did something weird?
You are never required to catch an exception.

Possibly you also tried to add a try block. Each try block requires at least one catch block. However, you do not need a try block to throw exceptions.
If you are not going to catch an exception, then don't add a try block.

---

Oh, ok. I didn't know that. I thought you did need a try block to throw exceptions.

Ok, it turns out that other error was completely un-related to the original error that I was getting. It was about something else entirely.

The last error that I mentioned was due to not defining a class function that I declared. I forgot I had declared that one, and it took me some time to figure out how to display the complete linking error in Qt Creator. Once I did, I immediately learned what was causing the error, and fixed it on my own.

Cheers...and thanks people. :)