Re: [boost] [1.56.0] Release candidates available

Subject: Re: [boost] [1.56.0] Release candidates available
From: Beman Dawes (bdawes_at_[hidden])
Date: 2014-08-02 08:36:32

On Wed, Jul 30, 2014 at 5:42 PM, Beman Dawes <bdawes_at_[hidden]> wrote:

>
> On Tue, Jul 29, 2014 at 6:33 PM, Eric Niebler <eniebler_at_[hidden]> wrote:
>
>> On 07/29/2014 02:44 PM, Boris Schäling wrote:
>> > On Tue, 29 Jul 2014 19:06:43 +0200, Marshall Clow
>> > <mclow.lists_at_[hidden]> wrote:
>> >
>> >> Release candidate files for 1.56.0 are available at
>> >> http://boost.cowic.de/rc/
>> >
>> > I get a compiler error with g++ 4.8.3 on Cygwin (64-bit):
>> >
>> > In file included from libs\system\src\error_code.cpp:19:0:
>> > ./boost/system/detail/error_code.ipp: In member function ‘virtual
>> > std::string
>> > boost::system::{anonymous}::system_error_category::message(int) const’:
>> > ./boost/system/detail/error_code.ipp:413:5: error:
>> > ‘local_free_on_destruction’ is not a member of ‘boost::detail’
>> >      detail::local_free_on_destruction lfod(lpMsgBuf);
>> >      ^
>> >
>> > This looks like the bug Eric had reported at
>> > <https://svn.boost.org/trac/boost/ticket/10137>.
>>
>> Crud. Beman, did a merge get missed?
>>
>
> It was a pull request. I've merged it to develop, although I only tested
> it on VC++ and MinGW because I'm having trouble getting Cygwin tests to
> work because of some flaw in my test setup. I'll take a break, then look at
> it again.
>
> But even if it is working for Cygwin, I'd want to see successful tests on
> Linux and OS X at a minimum before merging to master.
>

I've now tested Boost.System locally:

OS X / gcc
OS X / clang
Ubuntu / gcc
Ubuntu / clang
Win 7 / msvc-10.0
Win 7 / msvc-11.0
Win 7 / msvc-12.0
Win 7 / gcc - cygwin / cygwin installer / from bash (thanks to Eric for
help getting this running)
Win 7 / gcc - cygwin / cygwin installer / from Windows command line
Win 7 / gcc - mingw / mingw installer / from Windows command line

The last two turned up a different configuration issue which has also been
fixed.

I'd like permission to merge system develop to master Sunday morning (after
more regression testers have cycled)>

--Beman