- Boost - lists.boost.org

doubts about python3 support
by Neal Becker 26 Aug '13

26 Aug '13

I'm building using: bjam -d+2 -j8 cxxflags='-march=native -Wno-unused-local-typedefs' -sHAVE_ICU=1 - sEXPAT_INCLUDE=/usr -sEXPAT_LIBPATH=/usr/lib64 --layout=tagged threading=single,multi link=shared stage In user-config.jam I added: using python : 3.3 : /usr/bin/python3 : /usr/include/python3.3m ; I wind up with stage/lib/libboost_python.so stage/lib/libboost_python3.so But AFAICT, they were build the same way. It's be nice if one was built against python2 and the other python3, but that doesn't appear to be true. AFAICT, both were built from the same sources (using /usr/include/python3.3m). What I'd like is to build both python2 and python3 from the same source tree, installing the final shared libs into the same destination dir. Right now I have 2 different source trees, and build switching the one line in user- config.jam.

2 3

MSVC doesn't support "not"?!
by Daryle Walker 26 Aug '13

26 Aug '13

Just checked the regression testing pages. Most of the testing stations that reported fails for Boost.Rational have switched to pass since I've uploaded updates. Besides the stations that haven't run new tests, and the "teeks" team that run Clang tests although it seemingly isn't actually installed, the newly-run MSVC stations report fails. The old errors are gone, but the new error is: ..\libs\rational\test\rational_test.cpp(1012) : error C2065: 'not' : undeclared identifier ..\libs\rational\test\rational_test.cpp(1012) : error C2146: syntax error : missing ')' before identifier 'dummy' ..\libs\rational\test\rational_test.cpp(1012) : error C2059: syntax error : ')' This happens with MSVC 10.0 and 11.0. Hasn't "not" been an alternate symbol for "!" since C++98? I'll change it to "!", but that doesn't excuse not fixing it for 15 years! What happened? (I think some of the MSVC authors lurk here.) Daryle W.

6 7

[config] BOOST_NO_SFINAE_EXPR and GCC 4.6.4
by Agustín K-ballo Bergé 26 Aug '13

26 Aug '13

According to this table http://gcc.gnu.org/projects/cxx0x.html, GCC supports "Solving the SFINAE problem for expressions" since 4.4. According to Boost.Config, it is since 4.5. The following test case works fine with 4.8.1, but fails miserably with 4.6.4: http://ideone.com/RzgXKM Errors include invalid conversion from X const* to X*, and use of non-member-ptr in member-ptr expression. GCC 4.7.3 (tested by zao at #boost) shows similar results: https://www.acc.umu.se/~zao/kbsfinae11.txt Should BOOST_NO_SFINAE_EXPR be adjusted according to the results of this test case? Regards, -- Agustín K-ballo Bergé.- http://talesofcpp.fusionfenix.com

3 7

[RPC] Boost.Asio to write *concurrency ready* code
by microcai 26 Aug '13

26 Aug '13

Hi, boost experts! This is my first mail. Although I am not boost newbee, I am indeed new to the boost community. == I was using Boost.Asio for a long time, for *Async Network I/O*, but up till recently, I made an amazing discover that Boost.Asio is NOT ONLY a Network library, but indeed a very *general* toolkit for building concurrent algorithm. I made the first discover when my friend show up an concurrent sum ``` #include <iostream> #include <vector> #include <algorithm> #include <numeric> #include <future> template <typename RAIter> int parallel_sum(RAIter beg, RAIter end) { typename RAIter::difference_type len = end-beg; if(len < 1000) return std::accumulate(beg, end, 0); RAIter mid = beg + len/2; auto handle = std::async(std::launch::async, parallel_sum<RAIter>, mid, end); int sum = parallel_sum(beg, mid); return sum + handle.get(); } int main() { std::vector<int> v(10000, 1); std::cout << "The sum is " << parallel_sum(v.begin(), v.end()) << '\n'; } ``` Here the actual calculations are done async and concurrently. But, the problem remains: ** The total threads that this algorithm spawn is not predictable and not configurable. ** And that's why we will not like such algorithm altogether. Days later, when I am review an old code that was doing an *async directory walk through* , I soon discover an general *concurrency ready* algorithm that the number of threads utilized can be fully controlled. The code utilized *any number of threads that is doing io_service::run()* Here is the good old demo that uses recursive algorithm : ``` // this is sequential recursing version void hanoi_sync(int n, char A, char B, char C) { if (n == 1) { printf("Move disk %d from %c to %c\n", n, A, C); } else { hanoi_sync (n-1, A, C, B); printf("Move disk %d from %c to %c\n", n, A, C); hanoi_sync (n-1, B, A, C); } } ``` But with Boost.Asio, we turn it into parallelism algorithm: ``` // concurrent hannoi tower solver void hanoi_async(boost::asio::io_ service& io, int n, char A, char B, char C) { if (n == 1) { printf("thread id %d, Move disk %d from %c to %c\n", boost::this_thread::get_id(), n, A, C); } else { io.post(boost::bind(&hannoi_async, boost::ref(io), n-1, A, C, B)); printf("thread id %d, Move disk %d from %c to %c\n", boost::this_thread::get_id(), n, A, C); io.post(boost::bind(&hannoi_async, boost::ref(io), n-1, B, A, C)); } } ``` Thanks to the amazing io_service.post, we can now turing any (if not mostly) *recursing code* into *parallelism code*. And the threads that used by such *parallelism code* is fully controlled. If one thread is calling io_service::run(), then it is simple an async version of the sequential recursing one, if more that one thread is doing io_service::run(), then you know that the job is now run in parallel. And yet there needs only one place to control how may threads is used. Want it to be faster? spawn more threads to call io_service::run()! The code will *automatically* use all threads that you have allocated to Boost.Asio. I am *not sure* if anyone else has made such discovery, anyway, comment is welcomed.

1 0

[type_triats] [move] is_copy_constructible<T> trait
by Antony Polukhin 26 Aug '13

26 Aug '13

Hi, I have an implementation of is_copy_constructible<T> trait. It works great with C++11, in C++03 it detects constructors assuming that if type is derived from boost::noncopyable or uses BOOST_MOVABLE_BUT_NOT_COPYABLE macro - it has no copy constructor. Questions are: * May I add that trait to trunk version of Boost.TypeTraits (with documentation and tests, as usual)? * May I or Boost.Move maintainer apply this patch to BOOST_MOVE_IMPL_NO_COPY_CTOR_OR_ASSIGN in boost/move/core.hpp : #ifdef BOOST_NO_CXX11_DELETED_FUNCTIONS #define BOOST_MOVE_IMPL_NO_COPY_CTOR_OR_ASSIGN(TYPE) \ private:\ TYPE(TYPE &);\ TYPE& operator=(TYPE &);\ + public: \ + typedef int boost_move_no_copy_constructor_or_assign; \ + private: \ // #else Why do I need this trait? Well, it allows to implement move_if_noexcept() that is required for circular_buffer and possibly for other containers. -- Best regards, Antony Polukhin

2 4

[git] Some questions on Git transition
by Daryle Walker 26 Aug '13

26 Aug '13

* I see we have a GitHub account at https://github.com/boostorg, with repositories for each library. If we add a new library, and it has been Git-based, are we going to mirror it in the Boost account? That'll make sure if something happens to the original, we still have a last-known-good to work with. Would having all the libraries (or copies/mirrors) under the same account make scripts easier? What about non-Git libraries, will they be copied over? Would a bridge be used? (There's at least bridges for Git connecting to either Subversion or Mercurial.) * We currently ship all the libraries merged together as a single archive. Some package managers even provide the Boost libraries as a single item. When we go Git, is that still going to be in place (with the release managers doing some sort of merging script), or are we and every user going to move to doing several dozen downloads instead? (And everyone has to figure out how to use more than one Boost library at once.) * Are we eventually going to use something like http://nvie.com/posts/a-successful-git-branching-model/? If so, are we going to use that to script releases? (At release-time, grab the latest version of each library from their master branches, apply the merging script, run tests, then ship.) Daryle W.

2 3

[release] How do you make a library update official?
by Daryle Walker 24 Aug '13

24 Aug '13

I just uploaded changes ([85417] and [85418]) to Boost.Rational that fix the two major problems showing up on the regression tests. How do I get them on the official distribution? I know we don't read off the SVN-trunk directly; there's some staging area first. (That's why the shippers occasionally declare a code freeze deadline.) I know we have a "release" branch, but I'm not sure that's still being used. I want to blaze ahead making tweaks to Boost.Rational, but not if you can't "make official" a non-HEAD change-set. Even if I miss the deadline for these other tweaks, I want the two I just already did to be in Boost 1.55. Daryle W.

8 9

[admin] [svn] Boost.Commit list is not working
by Michel Morin 23 Aug '13

23 Aug '13

It seems that Boost.Commit mailing list has suspended for ten days. Regards, Michel

1 0

How not to use auto-link?
by Daryle Walker 23 Aug '13

23 Aug '13

Recently, I've been using CodeLite for Windows to do my Boost work. I have it running a MinGW-builds version of GCC 4.8 by default. I've enabled CodeLite to use Visual C++, since I have a copy of the Express version and want to avoid missing MSVC-related errors in the future. After telling CodeLite where MSVC and its directories are, I ran "rational_test.cpp" by itself, which gave up a linking error for Boost.Test. The linker was looking for those fancy named libraries that I didn't automatically make since I usually include all the source files. (And I'm using Boost from a Subversion download instead of a fixed one.) I added the source files from Boost.Test and #defined BOOST_ALL_NO_LIB=1 from the project options. The project-make system was still looking for library files! (I don't recall if it was asking for the same library file.) As I already built those files while experimenting with the documentation system, I added that directory to the search list and everything worked. I thought defining BOOST_ALL_NO_LIB was sufficient to disable auto-linking. What did I mess up? Daryle W.

2 1

[rational] Which way should std::numeric_limits::min go?
by Daryle Walker 23 Aug '13

23 Aug '13

The default implementation of a std::numeric_limits<T> instantiation is to have all static data member be set to static_cast<X>(0) and all the static member functions return T{}. When an instantiation is manually written, you're supposed to leave members for the types of numbers (integer vs. floating vs...) that you're NOT as in the unimplemented case. The static member function "min" has a defect. The function "denorm_min" returns the smallest positive value that can be represented (denormalized or not), and is implemented only for floating types. The function "lowest," new to C++11, returns the regular (i.e. not Infinity or NaN, etc.) value closest to negative-infinity. Lowest was introduced because for C++98/03, "min" returned "denorm_min" for the built-in floating types and "lowest" for built-in integer types! You can't get generic code out of that (without checking some more flags from numeric_limits). Which version of "min" should boost::rational take? I was think of "denorm_min," since I'll be implementing that. The built-in integer types use "lowest" for "min," but don't define the other one at all. Daryle W.

2 2