Private Static class members

Hi,

I'm trying to write a class to hold two variables to be used by a number of functions in the program. Here's the class:

Code:

---

class InBAF
{
private:
        static char location;
        static CString server;

public:
        static void SetLocation(char newloc)
        {
                location = newloc;
        }

        static void SetServerName(char location)
        {
                switch (location)
                {
                case 'T':       
                                // Input Toronto Server
                        server = "ftp.pureftpd.org";
                        break;

                case 'M':
                                // Input Montreal Server
                                server = "ftp.pureftpd.org";
                        break;
                }
        }

        static CString GetServerName()
        {
                return server;
        }
};
char InBAF::location = 'N';
CString InBAF::server = '\0';

---

when I try to compile I get this error for each member of the class:

Code:

---

private: static char  InBAF::location" (?location@InBAF@@0DA)
 already defined in GetData.obj

---

GetData is a different function in my project; it has nothing to do with this class.

What am I missing?

and I tried:

Code:

---

CString InBAF::server[0] = '\0';

---

but this comes up:

Code:

---

error C2466: cannot allocate an array of constant size 0
 error C2040: 'private: static class CString  InBAF::server' : 'class CString []' differs in levels of indirection from 'class CString'

---

I declare the class in the main header file. GetData.cpp has that header file included, and alphabetically it is compiled first. Can that be the problem? Should I define the class in a .cpp file?

Edit: I moved the class definition to the main .cpp file and everythign seems to be happy. But that doesn't really make sense to me....

>> CString InBAF::server = '\0';

You should not assign a character to a CString. If you want the CString to be empty, just don't initialize it when you define it (the default constructor will initialize it to be empty).

Code:

---

CString InBAF::server;

---

As with your other thread, you should never assign '\0' to the CString. A CString is a class that is not necessarily null-terminated, so assigning a null character to it doesn't make it empty. Always assign a string literal instead of a character literal to a CString (e.g. ""), or call the Empty() function to empty the contents.

>> or call the Empty() function
I didn't know about that function, but that's what I was trying to do with '\0'
and a string literal is " "
and a character literal is ' ', right? I haven't heard of it before.

Eh?

Unless you know the internal representation of strings in CString, you cannot assume anything about them. And, if you do know something about the internal representation, then you should promptly put it out of mind and try to forget it.

The \0 character is simply a signalling device -- a convention used with C-style strings. Instead of explicitly passing the string length everywhere (as is done with normal arrays), you simply read up until the \0. Now, a class may store an unterminated character array, and the length (which has the benefit of making calls to get the length constant time instead of linear time). If it does have a terminating character, then it will add it itself. So, when copying a char*, it will do what it has to with the terminating \0.

Yes, the double quotes are used for string literals, and the single quotes are used for character literals. So even if your string only has only zero or one characters, you would use double quotes if you want to treat it as a string. And of course a character literal in single quotes must be exactly one character.

OK, in sight of all this. I'm readign a filepath that needs to be used by a different function. But the read may use \ or / for directory separation. to pen a file I have to use either \\ or /. would this be appropriate:

Code:

---

for(int i=0; i<strlen(fname); i++)
                {
                        if(fname[i] == '\\')
                                fname.SetAt(i, '/');
                }

---

fname is a CString.

I would use the CString Replace function:

Code:

---

fname.Replace('\\', '/');

---

Additionally, I'd imagine that there isn't an overload of strlen for CString, but it has it's own length function.

(Though, the existence of a replace function definitely makes it the proper tool for this job.)

GetLength() is the proper length function for a CString, although technically strlen would work because CString has an implicit conversion to LPCTSTR which ends up being a const char*.

Coming back to the original problem, move the definition of the statics;

Code:

---

char InBAF::location = 'N';
CString InBAF::server = '\0';    // check this too, although it's not relevant to the original problem

---

out of the header file and into one (and only one) of the source files that #include's that header.

The reason the LINKER is complaining is that, by placing those lines into the header file, every source file that #include's that header instantiates a copy of the statics. That is a violation of the "One Definition Rule".

Couldn't he also use the code

Code:

---

#ifndef INBAF_H
#define INBAF_H

class InBAF
{
.
.
.
};
#endif

---

This is how I always do my classes is this bad form or something?

@~Kyo~:
No include-guards do nor help in this case.
The problem in this case is not that the .h-file is included multiple times in one compilition unit ( this is what include-guards are used for ) the problem is that this .h-file is included in different compilation units. So the variables InBAF::location and InBAF::server end up in different .o -files and the linker cannot resolve this conflict.
Kurt

I didn't do my homework properly in researching the CString functions.

Thanks guys,

AS.