Re: [boost] boost.config was Re: License of endian and limits in Boost detail

Subject: Re: [boost] boost.config was Re: License of endian and limits in Boost detail
From: Philip Bennefall (philip_at_[hidden])
Date: 2013-03-06 09:01:08

----- Original Message ----- 
From: "Daniel James" <daniel_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Wednesday, March 06, 2013 2:45 PM
Subject: Re: [boost] boost.config was Re: License of endian and limits in 
Boost detail

On 6 March 2013 13:07, Andrey Semashev <andrey.semashev_at_[hidden]> wrote:
> I wonder
> if we could simply drop detail/limits.hpp (and always rely on
> compiler-provided numeric limits)

boost/detail/limits.hpp is almost never used nowadays. It's only
included from boost/limits.hpp, using:

#ifdef BOOST_NO_LIMITS
# include <boost/detail/limits.hpp>
#else
# include <limits>
#endif

BOOST_NO_LIMITS is only defined in 'boost/config/stdlib/sgi.hpp'. Does
anyone still use that configuration? I suppose it might also be
defined in a custom configuration or by a user. It's quite likely that
the header is incompatible with other STL implementations anyway. So I
suspect it's only effectively used in code that is already subject to
the SGI license (i.e. the header is only included when already using
the SGI libraries), although I don't know if that makes a difference
legally.

For my part I want to distribute a small subset of Boost with my program, 
which I am also placing under the Boost Software license. But if either of 
these files get included, detail/limits and/or detail/endian, we get this 
uncertainty as to what really applies legally. If the file is hardly ever 
used, does it need to be kept if it causes legal ambiguities?

Philip Bennefall
_______________________________________________
Unsubscribe & other changes: 
http://lists.boost.org/mailman/listinfo.cgi/boost