Boost C++ Libraries: Ticket #3998: Provide light weight headers for extensibility features of BOOST_FOREACH

attachment set

Thu, 11 Mar 2010 15:18:54 GMT

attachment → foreach_separate_headers_for_extensibility.patch

svn diff for trunk r60445

status changed

Eric Niebler — Wed, 09 Jun 2010 14:51:16 GMT

status new → assigned

Although I don't see a need to put the implementation of is_lightweight_proxy and is_noncopyable in their own header, putting forward declarations for them in a separate header is a good suggestion.

status changed; resolution set

Eric Niebler — Wed, 09 Jun 2010 14:52:56 GMT

status assigned → closed
resolution → fixed

(In [62661]) add foreach_fwd.hpp for forward declarations of foreach's customization points, fixes #3998

status changed; resolution deleted

Thu, 24 Jun 2010 21:51:44 GMT

status closed → reopened
resolution fixed

Although I don't see a need to put the implementation of ...

I'm sorry to have forgotten to put a link, but my another patch #3996 (for ptr_container) depends on the separation.

Could you please re-evaluate the separation?

Eric Niebler — Thu, 24 Jun 2010 21:57:24 GMT

What exactly is the problem you're experiencing? It's perfectly legal to specialize a template that has only been forward-declared, so the foreach_fwd.hpp should be all you need. If it's not, can you tell me what compiler/version you're using, send me the error message you're getting and code I can use to reproduce the problem? Thanks.

Eric Niebler — Thu, 24 Jun 2010 22:11:33 GMT

I looked more closely at your patch for #3996 and see the problem. For every instantiation of reversible_ptr_container, you are instantiating foreach::is_noncopyable to (optionally) inherit from noncopyable. This is the wrong way to do it and will lead to ODR violations when different translation units see different specializations of foreach::is_noncopyable. Instead, define a partial specialization of foreach::is_noncopyable< reversible_ptr_container<...> > as you first suggested.

status changed; resolution set

Eric Niebler — Thu, 24 Jun 2010 22:12:25 GMT

status reopened → closed
resolution → wontfix