Re: [boost] random and detail namespace

From: Beman Dawes (beman_at_[hidden])
Date: 2000-06-29 09:08:13

Jens Maurer wrote:

 > ...
 >
 >Howard Hinnant suggested a namespace random within namespace boost
 >where all definitions reside, possibly nested in deeper namespaces.
 >A "using namespace random" within namespace boost complements
 >the setup.  However, Beman said in a private e-mail that he
 >doesn't like that setup, although we have a precedent in
 >cast.hpp.

In cast.hpp, it seemed the pain outweighed the gain.  The pain included 
increased complexity and difficulty with broken compilers.

With larger libraries like random, the advantages seem to outweigh the 
disadvantages.  The B-tree library I am working on also has a similar 
problem, with similar advantages.

 >All things considered, it might be a good idea to move the
 >generator template definitions from namespace boost::detail into
 >namespace boost::random without a "using namespace random" in
 >namespace boost and adjust the documentation accordingly.
 >Other things, including the typdef's intended for user type (1)
 >and (1a) will remain where they are (i.e. namespace boost).

Yes, that sounds sensible.  If you decide to do that, please report your 
experiences, and then let's try to update boost programming policies and 
guidelines to match.

--Beman