Bruce Perens writes ("Bug#2447: dselect" compilation buglet"):
> I'm having trouble even with this patch installed. The problem goes away
> if I remove the "static" declaration from the functions. It looks as if it's
> a compiler bug.Yes, this is a compiler bug. Ray Dassen first noticed it in July, and I assumed that it would be fixed. I'm disappointed too see that it hasn't been fixed yet. (I haven't yet compiled dpkg as ELF myself, so I hadn't seen the problem.) When I next work on dpkg I shall write an autoconf macro to detect the problem and compile the relevant bits without optimisation (which makes the problem go away). > I think, however, that a function pointer that has been declared in the > "C" universe should not point to a function that is declared in the C++ > universe. I don't know if g++ will ever care about this, but I've met > other compilers that do. You're right, I shall change this (but it's not what's causing problems here - g++ doesn't mind about this at the moment). Ian.
