|
Boost Users : |
From: Bill Lear (rael_at_[hidden])
Date: 2005-01-07 14:54:05
On Friday, January 7, 2005 at 10:58:13 (-0800) Robert Ramey writes:
>...
>The whole concept behind BOOST_CLASS_EXPORT has been problematic for me for
>a couple of reasons. As a result, the confusion you cite is a real one.
I settled on the following that a friend suggested, which seems to
work.
In my serialization/util.h file, I have the following:
template <class T>
void save(const T* t, const std::string& filename) { ... }
template <class T>
T* load(const std::string& filename) { ... }
at the bottom of this file, I have the following:
namespace dummy { struct bogus { static int bogus_method(); }; }
// force linking with util.cc
static int bogus_variable = dummy::bogus::bogus_method();
in serialization/util.cc, I have:
#include <serialization/util.h>
// expensive include which we wish to relegate to one one source file.
#include <AllDerived.h>
// the trigger method that forces others to link with this module
int dummy::bogus::bogus_method() { return 0;}
The AllDerived.h file has includes for all my headers and all the
BOOST_CLASS_EXPORT macros.
In my application code, I do this:
#include <Base.h>
#include <serialization/util.h>
and it all works flawlessly. I am spared from including the headers
for my derived classes, and the accompanying export macros. The save
and load routines work as expected.
Bill
Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net