Re: [boost] C++03 / C++11 compatibility question for compiled libraries

Subject: Re: [boost] C++03 / C++11 compatibility question for compiled libraries
From: Andrey Semashev (andrey.semashev_at_[hidden])
Date: 2018-02-10 17:00:30

On 02/10/18 18:43, Peter Dimov via Boost wrote:
> 
> But, if you really wanted to know, libstdc++ in a situation like the 
> above (when the function has to be virtual because it's f.ex. specified 
> that way by the standard) does something like the following:
> 
> #if cpp11
> 
> virtual int __f( boost::function<int ()> const& F ) = 0;
> virtual int f( std::function<int ()> const& F ) = 0;
> 
> #else
> 
> virtual int f( boost::function<int ()> const& F ) = 0;
> virtual int __f( std::function<int ()> const& F ) = 0;
> 
> #endif
> 
> except of course we don't have std::function in C++03 mode, so even 
> declaring __f will be a problem.

If you rely on the particular algorithm of vftable slot allocation, 
declaring is not a problem - you can declare pretty much whatever 
virtual function. It's not going to be called anyway and its sole 
purpose is to reserve a particular slot.

libstdc++, though, has the luxury of knowing the slot allocation 
algorithm, as well as binary layout of its own components (e.g. 
std::string vs. std::__cxx11::string), so it can do stuff like that and 
even better (e.g. it can actually transparently support both string 
types, where needed). In an external generic library like Boost, this is 
a pretty risky thing to do. Just always using ABI-stable types in the 
link interface seems more preferable.