...

preparing boost.lockfree for trunk, i need a good way to figure out if the <atomic> header is available. would it be possible to add a macro to detect that? BOOST_NO_CXX11_HDR_ATOMIC or the like?

For sure, not sure why it got missed out to be honest, do you have a patch?

unfortunately no patch ... also not sure if it should go under config/stdlib or under config/compiler tim

...

preparing boost.lockfree for trunk, i need a good way to figure out if the <atomic> header is available. would it be possible to add a macro to detect that? BOOST_NO_CXX11_HDR_ATOMIC or the like?

thanks, tim

FYI, gcc 4.4 has <cstdatomic> instead of <atomic> with the same contents. I don't know about other compilers, maybe there are other similar cases. So we may be better have <boost/atomic.hpp> which defines the macro or includes the appropriate compiler header otherwise. Or perhaps this is the Boost.Atomic job.

one point is to have an <atomic> header, the second point is an compliant implementation of atomic<>. gcc did not support atomic<> for structs before 4.7 :/ tim

...

...

...

preparing boost.lockfree for trunk, i need a good way to figure out if the <atomic> header is available. would it be possible to add a macro to detect that? BOOST_NO_CXX11_HDR_ATOMIC or the like?

thanks, tim

FYI, gcc 4.4 has <cstdatomic> instead of <atomic> with the same contents. I don't know about other compilers, maybe there are other similar cases. So we may be better have <boost/atomic.hpp> which defines the macro or includes the appropriate compiler header otherwise. Or perhaps this is the Boost.Atomic job.

one point is to have an <atomic> header, the second point is an compliant implementation of atomic<>. gcc did not support atomic<> for structs before 4.7 :/

I guess, it means we need another macro then.

or BOOST_NO_CXX11_HDR_ATOMIC should be defined, unless the implementation is complete.