Subject: Re: [boost] [clang] Windows toolset support?
From: Joshua Boyce (raptorfactor_at_[hidden])
Date: 2012-11-06 14:12:42
On Wed, Nov 7, 2012 at 2:32 AM, Paul A. Bristow <pbristow_at_[hidden]>wrote:
> For me Boost_root is boost-trunk
> and I'm using the same Ruben's downloads I believe so I've
> set MINGW32_CLANG=c:/mingw32-dw2
> (And this download works OK with NetBeans - for hello world programs (I'll
> try building boost.test
> on that IDE next).
> I've clear out my user-config.jam for clang and tried to use your command
> line thus
> b2 -q -j %NUMBER_OF_PROCESSORS% toolset=clang cxxflags=-std=c++11
> address-model=32 architecture=x86
> --stagedir=stage/clang-x86 link=static runtime-link=static threading=multi
> define=WINVER=_WIN32_WINNT_WIN7 define=_WIN32_WINNT=_WIN32_WINNT_WIN7
> stage --with-test >
> build_dll_libs_trunk_test_JB_%date%_%tim%.log 2>&1
> which looks the same to me as yours, but trying to compile Boost.test
> (which I presume you also did)
> I get
> In file included from libs\test\src\cpp_main.cpp:16:
> In file included from .\boost/test/impl/cpp_main.ipp:20:
> In file included from .\boost/test/execution_monitor.hpp:81:
> In file included from .\boost/detail/fenv.hpp:69:
> error: no member named
> 'fenv_t' in the global namespace
> using ::fenv_t;
> error: no member named
> 'fexcept_t' in the global namespace
> using ::fexcept_t;
> It was to overcome this that I added other options (and then I get
> problems finding cstddef. But
> I'm puzzled how your system works without - and how you don't seem to need
> to define tool clang?
> Thanks for your help so far - I'm sure it will be *really* useful to get
> many people able to use
> this compiler.
> But any other suggestions?
> Unsubscribe & other changes:
It's great to finally see more interest for getting Clang on Windows
working with Boost.
As for your specific issues, I'm not quite sure what could be causing
that... I have been successfully compiling and using Boost.Test for my
I know that recently there have been some big changes on trunk to
Boost.Test which have caused crashes under Clang on some platforms (which
included Windows when I tested it), so yesterday (or the day before, I
forget) I switched to the release branch temporarily (I couldn't be
bothered finding the offending commit at the time as I was busy fixing
something else) and had no issues compiling there... Perhaps you could try
the release branch and see if you're having the same issues? If so, your
issues may be caused by the recent unstable Boost.Test change(s)...
If that's not it though, please let me know and I'll set up a VM with a
clean environment and try compiling under that to make sure there's nothing
I've forgotten about (as I probably have!), then I'll post a full list of
steps and scripts I used so others can reproduce my results. It will take
me a few days to do this though as I am going away today and won't have
access to my machine for 2 days.
Thanks, and good luck!
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk