- Boost - lists.preview.boost.org

mpi compile bug
by 陈立红 03 Sep '12

03 Sep '12

Hi guys, I have installed the OpenMPI_v1.6.1 and want to make use of boost mpi. But I cannot compile it with msvc10.0 successfully. The error message is like below. I found similar bug Ticket #5465 (new Bugs), but there's no one comment on it. Please help. E:\soft\boost_1_49_0>bjam stage --toolset=msvc-9.0 --with-mpi --stagedir="E:\sof t\boost_1_49_0\Debug\lib10" link=static runtime-link=shared runtime-link=static ָ��줿�ѥ��Ҋ�Ĥ��ޤ�� ָ��줿�ѥ��Ҋ�Ĥ��ޤ�� ָ��줿�ѥ��Ҋ�Ĥ��ޤ�� ָ��줿�ѥ��Ҋ�Ĥ��ޤ�� MPI auto-detection failed: unknown wrapper compiler mpic++ Please report this error to the Boost mailing list: http://www.boost.org You will need to manually configure MPI support. warning: skipping optional Message Passing Interface (MPI) library. note: to enable MPI support, add "using mpi ;" to user-config.jam. note: to suppress this message, pass "--without-mpi" to bjam. note: otherwise, you can safely ignore this message. Component configuration: - chrono : not building - date_time : not building - exception : not building - filesystem : not building - graph : not building - graph_parallel : not building - iostreams : not building - locale : not building - math : not building - mpi : building - program_options : not building - python : not building - random : not building - regex : not building - serialization : not building - signals : not building - system : not building - test : not building - thread : not building - timer : not building - wave : not building ...found 1 target... Regards, Lihong Chen

1 0

[trac] Library components: ATTENTION LIBRARY AUTHORS
by Rene Rivera 03 Sep '12

03 Sep '12

If you are the author of an accepted library please remember to ask to add the library to the trac bug system. It's important that others have a component to assign to bugs. In order to add the library component we need to know the trac *account* name you are using. Please check in trac to see if your library has a component (with this report <https://svn.boost.org/trac/boost/report/15>). If not, or if there are errors, please reply here with the new, or corrected, information. And if you don't happen to have a trac account please ask for one (<https://svn.boost.org/trac/boost/wiki#GettingaTracSVNUserid>). Thank you. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org (msn) - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim,yahoo,skype,efnet,gmail

2 2

[lockfree] Faulty (?) static assert
by Hartmut Kaiser 03 Sep '12

03 Sep '12

Hey, The lockfree::queue constructors raise a static assert if no capacity is specified. This contradicts the member initialization code close to the assert, for instance: queue(void): head_(tagged_node_handle(0, 0)), tail_(tagged_node_handle(0, 0)), pool(node_allocator(), has_capacity ? capacity : 0) { BOOST_STATIC_ASSERT(has_capacity); initialize(); } which implies that 'has_capacity' is allowed to be false. From what I can see 'has_capacity' could be false, indeed, in particular if the queue is not bounded in size. Would changing this to queue(void): head_(tagged_node_handle(0, 0)), tail_(tagged_node_handle(0, 0)), pool(node_allocator(), has_capacity ? capacity : 0) { BOOST_STATIC_ASSERT(has_capacity || capacity); initialize(); } reflect the actual intention? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 8

[lockfree] Type conversion warnings
by Hartmut Kaiser 03 Sep '12

03 Sep '12

Hey all, The attached patch fixes a couple of warnings about possible loss of precision caused by integral type conversions for MSVC. Ok to commit? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 1

[lockfree] freelist_stack::allocate and freelist_stack::deallocate should not be private
by Hartmut Kaiser 03 Sep '12

03 Sep '12

Hey all, Currently, the allocate/deallocate functions et.al. in lockfree::detail::freelist_stack are private. This inhibits reusing this class for implementing other lock-free datastructures as it's not possible to allocate anything which has not a constructor taking 0, 1, or 2 arguments (those are supported explicitly by exposing overloads for freelist_stack::construct). Would it be possible to make the mentioned functions 'protected' instead? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 3

Please create a category for atomic, context, and lockfree in trac
by Hartmut Kaiser 02 Sep '12

02 Sep '12

Please create a category for atomic, context, and lockfree in trac. Thanks! Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

3 4

[Review:Contract] Some questions
by Vicente J. Botet Escriba 02 Sep '12

02 Sep '12

Hi, First of all, thanks Lorenzo for your work on pre-processor programming. With this library, you have showed to me how far the pre-processing can go in terms of expressiveness. I have some questions before doing a complete the review: = C++11 compatibility = * Does the library use c++11 features when available? I was thinking e.g. on the emulated virtual specifiers. BTW, I think the "new" virtual specifier was removed from the standard, but maybe I'm wrong. * I think the library don't allow to declare C++11 specific classes/function. E.g * variadic templates * noexcept functions * constexpr functions * class/function attributes BTW, I have not see any example using exception specifications using 'throws'. Does the library support them? Could you confirm? and if not, could you tell us if you have a plan to support them? = Postconditions result = "Postconditions can access the function return value by declaring a variable of type |auto| and assigning it to the |return| keyword (the variable name is arbitrary but |result| is often used)." What prevent the library to the use of return instead of an auto variable postcondition( // Postconditions. auto old_value = CONTRACT_OLDOF value, // Old value(s). value == old_value + 1, // Assertion(s)... return == old_value ) instead of postcondition( // Postconditions. auto result = return, // Result value. auto old_value = CONTRACT_OLDOF value, // Old value(s). value == old_value + 1, // Assertion(s)... result == old_value ) =Constant assertions = I would expect that the old are always const so the following postcondition( auto old_even = CONTRACT_OLDOF even, auto old_odd = CONTRACT_OLDOF odd, // `[old_]even` and `[old_]odd` all `const&` within assertions. const( even, old_even ) even == old_even + 2, const( odd, old_odd ) odd == old_odd + 2 ) should be rewritten without loss of semantics as postcondition( auto old_even = CONTRACT_OLDOF even, auto old_odd = CONTRACT_OLDOF odd, // `[old_]even` and `[old_]odd` all `const&` within assertions. const( even ) even == old_even + 2, const( odd ) odd == old_odd + 2 ) = Oldof = As the old of value must be declared using a variable, using a specif oldof keyword that will replace auto old_size = CONTRACT_OLDOF size(), by oldofold_size = size(), could simplify the syntax. What do you think? Some minor points: * Why do you use( void ) instead of () in the functions without parameters? e.g. public: bool empty ( void ) const { return vector_.empty(); } Some suggestions: * Use 2 spaces indentation instead of 4 could reduce the width of the page, in particular when you use several columns. * The use of'int const' as return type is verbose and could be replaced by just int. Best, Vicente

2 21

[context] Recent changes (rev. 80363) break compilation
by Hartmut Kaiser 02 Sep '12

02 Sep '12

Hey all, The latest commit to trunk (rev. 80363) breaks compilation of Boost.Context on Windows. I'm attaching a patch fixing the issue. Ok to commit? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 2

[Review:Contract] Base class preconditions not checked
by Steven Watanabe 02 Sep '12

02 Sep '12

AMDG Test case attached. The problem is that the precondition test is executed when initializing a virtual base. Virtual bases are always initialized by the most derived class, so the preconditions of base class constructors are never checked. In Christ, Steven Watanabe

2 1

[lockfree] Inconsistent typedef tag_t in tagged_ptr
by Hartmut Kaiser 02 Sep '12

02 Sep '12

Hey all, The embedded typedef <...> tag_t has inconsistent visibility in the two implementations of tagged_ptr. In tagged_ptr.hpp this typedef is 'public', while in the file tagged_ptr_ptrcompression.hpp this typedef is 'private'. Would it be possible to make it public in both cases (patch attached)? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 2

[context] Windows ctx::maximum_stacksize() runtime asserts
by Hartmut Kaiser 02 Sep '12

02 Sep '12

Hey all, The Windows implementation of boost::ctx::maximum_stacksize() does always assert: std::size_t maximum_stacksize() { BOOST_ASSERT( ! is_stack_unbound() ); static std::size_t stacksize = 8 * 1024 * 1024; return stacksize; } That is because is_stack_unbound() will unconditionally return true. Patch is attached - ok to commit? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 2

MPI maintainer?
by Ralf Grosse-Kunstleve 02 Sep '12

02 Sep '12

Who is currently maintaining Boost MPI? We have two trac items I'd like to assign to the MPI maintainer. The first one has been sitting there for a long time already. https://svn.boost.org/trac/boost/ticket/6123 https://svn.boost.org/trac/boost/ticket/7316 Ralf

1 0

[test] Boost.Test unreleased since November 2009 ?
by Mathieu Champlon 02 Sep '12

02 Sep '12

Hi, I've noticed that no merge has been made from trunk to the release branch for Boost.Test since November 2009, meaning that a lot of bug fixes and features have remained unreleased. Is it on purpose ? Thanks, MAT.

2 1

[test] Where should I put replacement functions for WinCE?
by Jan Hudec 02 Sep '12

02 Sep '12

Hello Folks, Windows CE don't have time and clock functions, getenv function and errno.h (though they seem to have _errno function). I needed to define these things to get Boost.Test to build for WinCE. Now it would be nice to have this work by default. I can provide the definitions (they are quite simple), but I am not sure where would be correct place for them. Should I add them to config/platform/win32.hpp, create config/platform/wince.hpp or put them just in test? Thanks, Jan -- Jan 'Bulb' Hudec <bulb(a)ucw.cz>

2 1

[test] sigaltstack does not work on older Android
by Jan Hudec 02 Sep '12

02 Sep '12

Hello Folks, Android NDK contains several "platforms", depending on which release one desires to be compatible with. Boost test compiles fine with android-8 and newer, but the older versions have incomplete definition of sigaltstack (they define the structure, but not the function), causing compilation error in boost/test/detail/impl/execution_monitor.ipp. I have fixed it locally by adjusting the condition for defining BOOST_TEST_USE_ALT_STACK, but I am not sure which level of fix would be best suited for integrating in boost. I see 3 options: 1. Leaving it as is and simply saying android is only supported from API level 8 (that's 2.2; we are still compiling for 2.1 (which has no changes in native since 2.0)) 2. Simply disabling BOOST_TEST_USE_ALT_STACK in Android; it's not necessary after all. 3. Being really precise and reading the API version, but that requires including extra header. It would be nice, but I am not sure what is the correct place to include it. Here is a simple patch that just disables sigaltstack for Android: --- a/boost/test/impl/execution_monitor.ipp +++ b/boost/test/impl/execution_monitor.ipp @@ -173,7 +173,7 @@ namespace { void _set_se_translator( void* ) {} } # endif # endif -# if !defined(__CYGWIN__) && !defined(__QNXNTO__) +# if !defined(__CYGWIN__) && !defined(__QNXNTO__) && !defined(__ANDROID__) # define BOOST_TEST_USE_ALT_STACK # endif And here is a more complex patch that on android (the __ANDROID__ macro is predefined by the compiler in Android NDK) reads the api-level and uses it to decide: --- a/boost/test/impl/execution_monitor.ipp +++ b/boost/test/impl/execution_monitor.ipp @@ -173,7 +173,11 @@ namespace { void _set_se_translator( void* ) {} } # endif # endif -# if !defined(__CYGWIN__) && !defined(__QNXNTO__) +# if defined(__ANDROID__) +# include <android/api-level.h> +# endif + +# if !defined(__CYGWIN__) && !defined(__QNXNTO__) && (!defined(__ANDROID__) || __ANDROID_API__ >= 8) # define BOOST_TEST_USE_ALT_STACK # endif -- Jan 'Bulb' Hudec <bulb(a)ucw.cz>

2 1

[test] Fix WinCE workaround
by Jan Hudec 02 Sep '12

02 Sep '12

Hello Folks, boost/test/impl/execution_monitor.ipp has workaround for older MSVC++ and assumes WinCE version need it too. However the WinCE compiler caught up in next version. Here is a proposed patch: --- a/boost/test/impl/execution_monitor.ipp +++ b/boost/test/impl/execution_monitor.ipp @@ -84,7 +84,7 @@ using std::va_list; typedef unsigned uintptr_t; # endif -# if BOOST_WORKAROUND(_MSC_VER, < 1300 ) || defined(UNDER_CE) +# if BOOST_WORKAROUND(_MSC_VER, < 1300 ) || (defined(UNDER_CE) && BOOST_WORKAROUND(_MSC_VER, < 1500 )) typedef void* uintptr_t; # endif -- Jan 'Bulb' Hudec <bulb(a)ucw.cz>

2 1

[test] Interest in gtest-like output?
by Jared Grubb 02 Sep '12

02 Sep '12

I really like the Boost.Test framework, however, I cant help but have a bit of envy over how great the output from googletest looks. I've been playing around with writing a formatter/outputter that emulates the output style of googletest, and I am willing to spend some time to pretty it up and generate a patch if there's interest. Does anyone think this would be a worthy submission to boost? For an example of what the output looks like, this site has some great examples: http://www.ibm.com/developerworks/aix/library/au-googletestingframework.html (just to clarify: I'm not suggesting changing any BOOST_TEST syntax, just adding a new optional way to format the test progress and results to stdout) Jared

3 2

Re: [boost] [lockfree] Faulty (?) static assert
by Hartmut Kaiser 01 Sep '12

01 Sep '12

> The lockfree::queue constructors raise a static assert if no capacity is > specified. This contradicts the member initialization code close to the > assert, for instance: > > queue(void): > head_(tagged_node_handle(0, 0)), > tail_(tagged_node_handle(0, 0)), > pool(node_allocator(), has_capacity ? capacity : 0) > { > BOOST_STATIC_ASSERT(has_capacity); > initialize(); > } > > which implies that 'has_capacity' is allowed to be false. From what I can > see 'has_capacity' could be false, indeed, in particular if the queue is > not bounded in size. > > Would changing this to > > queue(void): > head_(tagged_node_handle(0, 0)), > tail_(tagged_node_handle(0, 0)), > pool(node_allocator(), has_capacity ? capacity : 0) > { > BOOST_STATIC_ASSERT(has_capacity || capacity); > initialize(); > } > > reflect the actual intention? Sorry, I meant: > BOOST_STATIC_ASSERT(has_capacity || !capacity); Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu > > Regards Hartmut > --------------- > http://boost-spirit.com > http://stellar.cct.lsu.edu > >

1 0

[atomic] Windows autolinking and longlong patches
by Hartmut Kaiser 01 Sep '12

01 Sep '12

Hey, Please find attached two patches for Boost.Atomic (trunk, rev. 80334) adding auto-linking for MSVC, fixing symbol visibility, and fixing a long long problem. Ok to commit? Regards Hartmut --------------- http://boost-spirit.com http://stellar.cct.lsu.edu

2 2

[type_eraseure]Why not make default concept_interface a metafunction returning Base?
by Larry Evans 31 Aug '12

31 Aug '12

Since the unspecialized concept_iterface definition: template<class Concept, class Base, class ID, class Enable = void> struct concept_interface : Base {}; does nothing, and it's used to compose the any base class heirarchy in: template<class Derived, class Concept, class T> struct compute_bases { typedef typename ::boost::mpl::fold< typename ::boost::type_erasure::detail::collect_concepts< Concept >::type, ::boost::type_erasure::any_base<Derived>, ::boost::type_erasure::concept_interface< ::boost::mpl::_2, ::boost::mpl::_1, T > >::type type; }; why not eliminate the empty concept_interfaces by redefining the unspecialized concept_interface to: template<class Concept, class Base, class ID, class Enable = void> struct concept_interface : Base { //The following enables mpl::fold, when using //this as the binary operator, to skip this //empty interface. typedef Base type; }; I believe that would work since mpl::fold uses the nested type if it's available, IIUC. Although it would not change the behaviour of any code, it would make the intent of the code clearer. Also, while trying to understand what the code is doing by using a type name demangler, I found all those empty concept_interfaces are just noise. -regards, Larry

1 0

[Review:Contract] Contract Broken Handlers , BOOST_NO_EXCEPT and noexcept
by Vicente J. Botet Escriba 31 Aug '12

31 Aug '12

Hi, I would like to know if Boost.Contract could work when BOOST_NO_EXCEPT is defined. I suspect that it can not at present, as the user needs to re-throw the exception to know the broken nature. Maybe the handler could have the an exception/error object as parameter instead. In order to be able to use contracts in C++11 functions declared noexcept I would need that the contract handler are declared noexcept also, in particular when the contract is broken from a destructor, which is often declared noexcept in C++11. Maybe the library can provide some specific set handlers that expect a noexcept handler for the destructor context. BTW, could the ALL CAPITAL |FROM_CONSTRUCTOR, ... enumeration literal|s be renamed to lower case from_constructor, ...? Best, Vicente

2 3

Question about release branches
by Pedro Larroy 30 Aug '12

30 Aug '12

Hi I'm not able to find some fixes such as (78513) in tags/release/Boost_1_51_0/ nor tags/release/Boost_1_50_0/ Why some fixes that are in trunk didn't made it to the release branches? Thanks. Pedro.

2 1

[range] Missing include in boost/range/algorithm/for_each.hpp
by Sergiu Dotenco 30 Aug '12

30 Aug '12

There is an issue with boost/range/algorithm/for_each.hpp which is missing an include. The problem seems to have been fixed in trunk two months ago (https://svn.boost.org/trac/boost/changeset/79180) but the changeset hasn't been applied to the release branch yet. Could this be done for the 1.52 release?

2 1

[fusion] undocumented sequence requirement: convert
by Eric Niebler 30 Aug '12

30 Aug '12

Say I have a valid fusion sequence. I want to use an mpl algorithm on it. I include boost/fusion/mpl.hpp and put a nested tag type in my sequence that is an alias for fusion::fusion_sequence_tag. So far, so good. Then I try to call mpl::pop_back on this sequence, and I find that, despite fulfilling all the documented requirements for Fusion ForwardSequence, the code fails to compile. That's because of this in boost/fusion/mpl/pop_back.hpp: template <> struct pop_back_impl<fusion::fusion_sequence_tag> { template <typename Sequence> struct apply { typedef typename fusion::result_of::pop_back<Sequence>::type result; typedef typename fusion::result_of::convert< typename fusion::detail::tag_of<Sequence>::type, result>::type type; }; }; My code doesn't compile because I haven't implemented the fusion::convert customization point. Where is that documented and what does it do? -- Eric Niebler BoostPro Computing http://www.boostpro.com

4 12

[type_erasure]bug in computed_bases?
by Larry Evans 29 Aug '12

29 Aug '12

Based on the output of the test code(compute_bases.cpp) in the 1st attachment, the compute_bases metafunction defined in any.hpp is producing something like: concept_interface < concepts , concept_interface < mpl::at_c<concepts,1>::type , concept_interface < mpl::at_c<concepts,2>::type , ID , Enable > , ID , Enable > , ID , Enable > where concepts is the Concept argument to any. When the test code is run with: //#define OVERLOAD_SIMPLE the output is what should be expected with a_any.simple(); calling the simple<.>::apply static method. The output for this case is shown in the compute_bases.not_overload_simple.out. OTOH, when the test code is run with: #define OVERLOAD_SIMPLE the code in the concept_interface<typename requirements<C>::type,...>::simple is run. The output for this case is shown in the compute_bases.yes_overload_simple.out. This output shows that the concept_interface<C,...> CTOR is called where C == mpl::vector<...>. Is this the intended behaviour? If so, where is this spelled out in the documentation? The documentation here: http://steven_watanabe.users.sourceforge.net/type_erasure/libs/type_erasure… certainly doesn't suggest anything of that sort, and the doc here: http://steven_watanabe.users.sourceforge.net/type_erasure/libs/type_erasure… says: 1. typename Concept The concept that we're specializing concept_interface for. which uses the singular Concept instead of the plural Concepts, suggesting the Concept is not an mpl::vector<> of any sort. -regards, Larry

2 4