On 10 August 2010 02:32, Josh Faust <jfaust@willowgarage.com> wrote:

Happy to hear that it works in general (would be pretty bad if it
didn't...). I'm more concerned about whether we're currently
specifying everything completely, or unknowingly exploiting
overlinkage.

I guess the only way to find out is to build and test everything with
that option included.

I just did two quick tests:
1) .so that overrides a global symbol (strstr in this case)

2) .so with a global object that has a constructor/destructor

#1 worked, #2 failed (the .so was removed).

If #1 failed I'd say we can't default to it. Not sure how important #2 is, but it's something that could definitely cause problems that would be difficult to track down.

Josh

Josh, how did you produce result 2)? I just tested it here to be aware of when it might break and couldn't reproduce the failure. It correctly excludes/includes the library when I comment/uncomment the relevant usage of the variable. Snippets below:

LIBRARY HPP:

namespace ecl {

class A {
public:
A() : i(3) {}
~A() {}
int i;
};
extern const A a;

}

LIBRARY CPP:

namespace ecl {

const A a;

}

MAIN CPP

int main() {

std::cout << "i: " << ecl::a.i << std::endl;

return 0;

}