Re: [Boost-users] 1.56.0 beta release candidate 2 available for testing

Subject: Re: [Boost-users] 1.56.0 beta release candidate 2 available for testing
From: Keith Bennett (kanezor_at_[hidden])
Date: 2014-07-15 19:22:55

On Tue, Jul 15, 2014 at 11:58 AM, Marshall Clow <mclow.lists_at_[hidden]>
wrote:

> The second release candidate files for 1.56.0 beta are available at
> http://boost.cowic.de/rc/
>
> As always, the release managers would appreciate it if you download
> the candidate of your choice and give building it a try. Please report
> both success and failure, and anything else that is noteworthy.
>
> This helps ensure the candidates build OK before we push them out to
> SourceForge.
>
> The files (and associated md5s) are:
> MD5 (boost-1_56_0_b1_rc2.7z) = cd6267325a8dadbb82b384c4776ee450
> MD5 (boost-1_56_0_b1_rc2.tar.bz2) = b74cd72043b6b2fa16c6690633acb422
> MD5 (boost-1_56_0_b1_rc2.tar.gz) = b51810be6737e73a4deee70e84895ca7
> MD5 (boost-1_56_0_b1_rc2.zip) = f4ceb73492ae7d9b1843d841498850e5
>
> Since this is the first post-git release, we are expecting some teething
> pains with this release.
> Please test extensively.
>
> Besides bug fixes the TypeIndex library has been included, and I think
> that the problems that mjklaim_at_[hidden] was seeing have been fixed.
>
> I would like to post this to SourceForge Thursday night/Friday morning, if
> possible.
>
> Thank you.
>
> --The release managers
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>

This appears to have built successfully for me on Ubuntu 14.04 using GCC
4.8.2. It appears to also have been built successfully on OS X using
clang-503.0.40. Both have tons of warnings. It appears most of the warnings
are from empty macro arguments while expanding boost.pp and boost.mpl
macros clang, "warning: empty macro arguments are a C99 feature
[-Wc99-extensions]". There are of course other warnings, but the vast
majority is that one. On GCC, the warnings aren't so bad...
mostly -Wunused-local-typedefs in boost.python and boost.spirit.

However, one of my projects now fails to build using boost 1.56, where
before it would succeed on boost 1.55. Further, it appears to only fail
when using OSX/clang: Ubuntu/GCC succeeds (well, after correcting for a
deprecated header). It appears to only fail when using --std=c++11 (which
my project does). If I omit --std=c++11, then the example compiles with one
warning.

I am not 100% certain if it's a fault in boost 1.56, as locally I am using
boost 1.55 from macports. When trying to check boost 1.55 "fresh", I cannot
seem to get boost.thread 1.55 to build on OSX/clang, which boost.log
requires, so cannot be 100% certain if this is a regression.

The issue I am seeing can be reproduced in a boost.log example:

Commands:
cd ~/Downloads
wget http://boost.cowic.de/rc/boost-1_56_0_b1_rc2.tar.bz2
tar -xjf boost-1_56_0_b1_rc2.tar.bz2
cd boost-1_56_0_b1
./bootstrap.sh
./b2 -j8 # see tons and tons of -Wc99-extensions warnings on OSX/clang
cd libs/log/example/settings_file
clang main.cpp -I /Users/kbennett/Downloads/boost-1_56_0_b1/ -L
/Users/kbennett/Downloads/boost-1_56_0_b1/stage/lib -l boost_log -l
boost_log_setup -l boost_date_time -l boost_filesystem -l boost_system -l
boost_locale -l boost_thread -DBOOST_ALL_DYN_LINK=1 -l stdc++ --std=c++11

Error Output:
In file included from main.cpp:25:
In file included from
/Users/kbennett/Downloads/boost-1_56_0_b1/boost/log/common.hpp:24:
In file included from
/Users/kbennett/Downloads/boost-1_56_0_b1/boost/log/sources/global_logger_storage.hpp:21:
/Users/kbennett/Downloads/boost-1_56_0_b1/boost/smart_ptr/make_shared_object.hpp:254:20:
error: no matching function for call to 'sp_forward'
    ::new( pv ) T( boost::detail::sp_forward<Args>( args )... );

-- 
Keith Bennett