compile time error...

when I use this:

Code:

---

string reason( "connect() failed" );
throw ( logic_error( reason ) );  // this is line 65.

---

I get the following compile-time errors:

Code:

---

C:\Ruchikar\RnD\wbsrvc1\Session.cpp(65) : error C2061: syntax error : identifier 'reason'
C:\Ruchikar\RnD\wbsrvc1\Session.cpp(65) : error C2066: cast to function type is illegal
C:\Ruchikar\RnD\wbsrvc1\Session.cpp(65) : error C2059: syntax error : ';'

---

But using this:

Code:

---

throw ( logic_error( string( "connect() failed." ) ) );

---

works just fine! What's the difference?

I'm using MS VC++ on Win2K.

Thanks in advance.

Your first 3 error codes are most likely trying to tell you it doesn't know what "string" is. You need to include <string.h> if you have not already done so, then everything should work.

not string.h, <string>

That said, it has me puzzled.

If the following is acceptable, then logic_error an take a string (or a type that can convert to one).

throw ( logic_error( string( "connect() failed." ) ) );

But, this then fails

string reason( "connect() failed" );
throw ( logic_error( reason ) ); // this is line 65.

Which is almost the same thing. logic_error is still getting construced with a string as a parameter.

Quote:

---

Originally posted by SilentStrike
not string.h, <string>

That said, it has me puzzled.

If the following is acceptable, then logic_error an take a string (or a type that can convert to one).

throw ( logic_error( string( "connect() failed." ) ) );

But, this then fails

string reason( "connect() failed" );
throw ( logic_error( reason ) ); // this is line 65.

Which is almost the same thing. logic_error is still getting construced with a string as a parameter.

---

Ah, silly namespaces ;) Yes, I always was puzzled on why the first way didn't work and an almost identical second did not. The first 3 errors though usually mean the header file was missing.

Maybe 'reason' goes out of scope. check for syntax errors immediately above line 65. If that doesnt work, you might have to post the code inbetween

<string> is included. That's why compiler didn't complain at line 64:

Code:

---

string reason( "connect() failed" );

---

The only errors I get are the ones in the original post.

I can't see any difference between the string object created in both the cases. If first one goes out of scope, so should second one also. Isn't it?

Can anybody try it with Borland and let me know if it is not some strange behaviour on MS VC++ part. Although, I think that the problem is with the code and not compiler.

The following compiled warningless with all warnings enabled on gcc 2.96 and 3.0.4.

Code:

---

#include <stdexcept>
#include <string>

using namespace std;

int main() {
        string reason( "connect() failed" );
        throw ( logic_error( reason ) );  // this is line 65.

        throw ( logic_error( string( "connect() failed." ) ) );

        return 0;
}

---

Well, the same code as SilentStrike's, when compiled on MS VC++ gave this:

Code:

---

Compiling...
test.c
c:\program files\microsoft visual studio\vc98\include\eh.h(32) : fatal error C1189: #error :  "eh.h is only for C++!"
Error executing cl.exe.

test.exe - 1 error(s), 0 warning(s)

---

Talk about being consistent. :rolleyes:

Is using MS VC++ not that good an idea?