Re: C++ Bindings Generate static const Code that will not Compile

[Bowie Owens <bowie owens csiro au>]

lgust verizon net wrote:

> Why does ORBit2 generated a static const char * data member of a class, which will not compile with gcc 3.2?
>  
It is a bug. I have applied a patch to in cvs. Can you please upgrade to 
the latest version of orbitcpp in CVS and confirm that the patch works 
for you.

> The C++ standard states that it is legal to initialize static const data members in a class that are enumerated or integral types.
>  
It is a shame they ruled out non-integral constants. Fortunately for 
lazy people like me GCC copes with non-integral constants in headers.

> The C++ Standard ISO/IEC 14882:1998(E) on page 158 states: “If a static data member is of const integral or const enumeration type, its declaration in the class definition can specify a constant-initializer which shall be an integral constant expression.”
>
> Since the generated statement is not of integral nor enumeration type, this statement generates the following errors when compiled by gcc 3.2:
>
> gcc -c -g -DORBIT2=1 -D_REENTRANT -I/home/nfr/local/include/orbit-2.0 -I/home/nfr/local/include/linc-1.0 -I/home/nfr/local/include/glib-2.0 -I/home/nfr/local/lib/glib-2.0/include -I/home/nfr/local/include/orbitcpp-2.0 client.cc
> In file included from cf-cpp-stubs.h:11,
>                 from client.cc:3:
> cf-cpp-common.h:46: invalid in-class initialization of static data member of 
>   non-integral type `const char*'
>  
For GCC (at least) the problem is not that the string is of non integral 
type, but that the string is not constant. The type should be const 
char* const rather than const char*. I have changed this in CVS.

> Is the syntax of the IDL file incorrect? If it is correct, then why does ORBit2 generated C++ code that will not compile?
>  
Your IDL was fine, so it is a bug in orbitcpp. If the idl file is 
syntactically correct, the backend should generate legal C++ code. Any 
time legit IDL generates bad C++ is pretty much going to be either a 
bug. Please continue to report problems as you find them.


--
Bowie Owens

CSIRO Mathematical & Information Sciences
phone  : +61 3 9545 8055
fax    : +61 3 9545 8080
mobile : 0425 729 875
email  : Bowie Owens csiro au