Re: [boost] Some statistics about the C++ 11/14 mandatoryBoostlibraries

Subject: Re: [boost] Some statistics about the C++ 11/14 mandatoryBoostlibraries
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2015-05-15 16:38:06

On 15 May 2015 at 20:29, Peter Dimov wrote:

> > Indeed, APIBind is there to ensure that your library X built against 
> > boost::function doesn't collide with your library X build against 
> > std::function in the same translation unit.
> 
> Does this actually work? Because I've trouble visualizing it based on this 
> description.
> 
> OK, X built against boost::function has
> 
> namespace boost { namespace foo {
> 
> template<class R, class... Args>
> using function = boost::function<R(Args...)>;
> 
> } }
> 
> and X built against std::function has
> 
> namespace boost { namespace foo {
> 
> template<class R, class... Args>
> using function = std::function<R(Args...)>;
> 
> } }

Correct.

> Why do those two not clash? I'm not seeing it. Perhaps the two 'foo' 
> namespaces are actually different?

Correct:

namespace boost { namespace foo { inline namespace ???_boost_??? {

template<class R, class... Args>
using function = boost::function<R(Args...)>;

} }

namespace boost { namespace foo { inline namespace ???_std_??? {

template<class R, class... Args>
using function = std::function<R(Args...)>;

} } }

The inline namespace is some mangled representation reflecting the 
different bind in the implementation.

In the above boost::foo::function will now conflict if used that way 
in the same TU. This is why under APIBind you namespace alias an 
exact version and mangling into your using namespace:

namespace user { namespace foo = ::boost::foo::???_std_???; }

Or as APIBind does it:

namespace user { namespace foo = BOOST_FOO_NAMESPACE; }

... as APIBind lets you reconfigure and reinclude the header for a 
library arbitrarily.

> But if so, does this not lead to a combinatorial explosion when more than 
> one component can use either its Boost or standard version? function, bind, 
> shared_ptr, regex... that's 16 combinations so far.

Correct. This was raised in Q&A in the talk too. I would suggest 
pruning the combinations in this case to a reasonable subset, though 
I have a python script which generates the macro boilerplate for you 
so you don't have to.

Niall

-- 
ned Productions Limited Consulting
http://www.nedproductions.biz/ 
http://ie.linkedin.com/in/nialldouglas/