Niall Douglas wrote
For dependency management, you're probably not going to like my solution as I don't either, but essentially you git submodule all your library dependencies into your include directory, so library/include/boost/library/submodule. You then go ahead and use them directly, so instead of:
#include "boost/spinlock/spinlock.hpp"
you do:
#include "spinlock/include/boost/spinlock/spinlock.hpp"
Well for once we do agree - I don't like it either! Right now we have all the boost library modules in a "flat" list at the same level of the library / application I'm working on. Rather than using b2 headers to create a monolithic head structure from our modules we could just as well replace #include <boost/spinlock/spinlock.hpp> with (assuming we're in side of libs/serialization/* (test, library or example) #include <../../spinlock/include/boost/spinlock/spinlock.hpp> or just add all the ../../*/include/ paths to the include search path. Then there is the redundant "boost" I'd like to eliminate - but much too disruptive for now. Personally, I'm at about the limit of git complexity I can currently deal with Robert Ramey -- View this message in context: http://boost.2283326.n4.nabble.com/List-of-C-11-only-Boost-libraries-and-the... Sent from the Boost - Dev mailing list archive at Nabble.com.