- Boost - lists.boost.org

[website] bcp link broken
by Eric Niebler 22 Nov '12

22 Nov '12

Go here: <http://www.boost.org/doc/tools.html>. Click on the link to bcp. Get taken nowhere. Any clues? TIA, -- Eric Niebler BoostPro Computing http://www.boostpro.com

4 4

[dynamic_bitset] expression template?
by Ben Coates 21 Nov '12

21 Nov '12

dynamic_bitset currently allocates temporary vectors for most operations, and all the allocating and deallocating can easily dominate the running time for a program that makes heavy use of them. You can work around some of it with careful use of named temporaries around loops but some operations like intersection of 3 sets ( (a & b & c).any() ) result in a lot of unwanted copies and allocations. I've had good results converting it to use expression templates for intermediary operations, with the outcome of being able to do arbitrary non-resizing operations without allocating. Is there any interest in changing the boost version to do this?

2 2

Should the clang, c++11 and libstdc++ configuration supported?
by Vicente J. Botet Escriba 21 Nov '12

21 Nov '12

Hi, clang compiler is not delivered with a C++11 standard library. When compiling on c++11 mode, some boost libraries (in particular, Boost.Thread, Boost.SmartPtr, Boost.Test) are expecting to have a C++11 standard library. This mean that the user needs to install the latest version of libc++. Could we state clearly in the release notes if the combination clang+ -std=c++11 + libstdc++ is not supported. Otherwise these libraries should update their code so that the user can use this combination. Best, Vicente

8 21

[stl_ext_adv] comparison with Boost.Container and string algorithms
by Vadim Stadnik 21 Nov '12

21 Nov '12

Hi all, In response to the previously made suggestion, I have added to the documentation the comparison of new proposed STL containers with Boost.Container. Please let me know if some of my statements concerning Boost.Container are incorrect, I will update the documentation. >From a user perspective an augmented array based B+ tree of the new library can be regarded as a segmented vector that offers several more efficient operations than std::vector: logarithmic time insertion, erasure, splice and split. One such data structure can support interfaces of all C++03 standard sequences and associative containers. This is why new containers offer a higher level of uniformity and generalization of the C++ standard containers. Unlike std::vector, the new array based containers do not have the problem of exponential storage growth and manual shrinking storage to the optimal capacity. The comparison section also includes simple tests of performance. One test shows the advantage of the wide set of efficient operations of new containers. None of the C++ standard and Boost containers can outperform new containers in the test with just two operations: move an iterator to a specified position and then insert a new element. The other test shows that new containers have good locality of reference. In sequential processing they perform much better than containers using dynamically allocated data structures. String algorithms The new sequence container that implements the union of interfaces of std::vector, std::list and std::deque is particularly beneficial for string algorithms. A container, such as std_ext_adv::sequence<char>, efficiently supports all main editing string operations: cut, copy, paste, delete, find, replace, etc. In terms of computational complexity this new container provides general and equally efficient support for both copy and move semantics operations. The move semantics functions splice() and split() can be used to replace inefficient copy semantics functions insert() and erase() to achieve significant performance improvements. The examples section and code shows how to concatenate two strings and extract a sub-string in logarithmic time using the move semantics functions. These operations do not require that user algorithms count moved characters. This functionality and efficiency are not possible with linked lists and basic search trees. The efficient calculation of statistical parameters and numerical integration in logarithmic time are the other practically important applications of the new containers. They are discussed in the examples section. Project (source code and documentation): https://github.com/vstadnik/stl_ext_adv_review Documentation: http://dl.dropbox.com/u/51041088/stl_ext_adv_review/doc/index.html Regards, Vadim Stadnik

1 0

Re: [boost] [Boost-build] [VxWorks] Updating boost/config/platform/vxworks.hpp for VxWorks 6.9
by Vicente J. Botet Escriba 20 Nov '12

20 Nov '12

Le 20/11/12 15:30, Brockamp, Peter a écrit : > Hello everybody, > > I'm new to the Boost.Build mailing list, so please forgive me any newbie > mistakes. I hope this is the right place for my post, if not please > redirect me to where it belongs! > I'm replied to the boost dev ML as most of the patches concern Boost.Config. > OK, some time ago I had to realize that the platform configuration for > VxWorks boost/config/platform/vxworks.hpp appears to be badly outdated > and doesn't work with the actual version 6.9. I started to modify this > platform header and after a while managed to get Boost compile and run > under VxWorks 6.9, at least as far as we use it here today. I could > offer to the community to update the existing header if this is desired > (I suppose it is). If so should I post the source code here or how to > proceed? Yep, a newbie question. You can start by creating a ticket on the Trac system https://svn.boost.org/trac/boost/newticket. You could add a patch associated with the ticket. > > My modifications surely aren't perfect/complete but it might be a fresh > start for this OS: > > - I have only tested it under RTP (Real Time Process) of VxWorks, not > DKM (Downloadable Kernel Module), due to lack of time. DKM has many > limitations compared to RTP, so this might cause problems! > > - Equally, due to time constraints, I tested GCC only, not the > diab-compiler, as we don't use it here. Same as above applies here, > problems might arise and I could not guarantee I might be able to fix > them. > > - And I tested x86 only, as we actually don't develop for ARM, etc. > > - I could test with V6.9 only, as we are not using the prior versions, > so maybe there will be other problems yet unknown. But at least I could > confirm that it is impossible to get Boost running under early versions > of VxWorks (V5.x and older). These versions are lacking a couple of > library functions required by Boost, they come with a library which is > partially non-ANSI-conforming (especially the STL) and the compiler (GCC > 2.96) is outdated beyond believe and lacks some functionality (mainly > template stuff) needed by Boost. So some OS-version checking scheme > would have to be added to vxworks.hpp, bailing out in case of obsolete, > unsupported versions. > > - A support guy from WRS (Wind River System) mentioned that they are > thinking of including Boost into their OS, so they might do some > adaption work in the future. But until now it doesn't appear to me > there's much activity from WRS, that's why I decided to offer what I > have done so far. > > - When compiling Boost for VxWorks I didn't use the build-system, as > VxWorks does a cross compile for the destination platform and I didn't > manage to figure out how to auto build this. Instead I manually pulled > all of Boost's files into a Workbench Project (Workbench is the eclipse > based IDE of VxWorks). So there would be some work left to get this up & > running as well. > > - My work is based mainly on modifying and supplementing most parts of > vxworks.hpp. Anyway, during the tests I did I stumbled upon some > problems in other files, e. g. caused by missing #ifdef's for VxWorks, > and even errors in a few of Boost's files, which I fixed as well. Should > all these fixes be posted here, or what's the right way to do it? > > > So, if you see some potential use in what I have done, despite the > incompleteness pictured above, please someone responsible mail me with > further instructions. > Best, Vicente

1 0

Boost.Thread 1.52 vs 1.51 - g++ 4.7.2
by Jose 20 Nov '12

20 Nov '12

When including thread.hpp (1.52) I get the error below. Same setup but Boost 1.51 works ok. Does anybody know what is the issue ? thanks # g++ -c test.cpp In file included from /usr/include/boost/thread.hpp:24:0, from test.cpp:1: /usr/include/boost/thread/future.hpp:139:26: error: expected ‘;’ at end of member declaration /usr/include/boost/thread/future.hpp:139:32: error: ‘BOOST_THREAD_NOEXCEPT_OR_THROW’ does not name a type /usr/include/boost/thread/future.hpp:139:19: error: looser throw specifier for ‘virtual const char* boost::future_error::what() const’ In file included from /usr/include/boost/thread/exceptions.hpp:21:0, from /usr/include/boost/thread/pthread/thread_data.hpp:10, from /usr/include/boost/thread/thread.hpp:17, from /usr/include/boost/thread.hpp:13, from test.cpp:1: /usr/lib/gcc/x86_64-redhat-linux/4.7.2/../../../../include/c++/4.7.2/stdexcept:70:5: error: overriding ‘virtual const char* std::logic_error::what() const throw ()’

2 1

Graph Algorithm Advice
by J Birch 20 Nov '12

20 Nov '12

This isn't a Boost Graph problem per se, just a graph algorithm question. Let's say I have the following graph: I want to represent this graph in a certain way. In the small legend box above, there are 2 versions of a correct answer. I want to print out each vertex followed by it's immediate adjacent vertex. If it has more than one, then each path from that point should be wrapped in a '()'. Currently, I am using the Ruby RGL package and have tried both a directed and an undirected graph using a depth first search traversal, and have yet to hit on an algorithm that takes every edge case. Anyone out there have some pointers on how to do this? Thanks! Birch

2 3

Failing to build Boost.Test library using Clang 3.1 on Netbeans
by Paul A. Bristow 20 Nov '12

20 Nov '12

Target: i686-w64-mingw32 Thread model: posix) (This is Ruben's mingw64 build) on Netbeans 7.2 on Windows 7. The compilation stage goes OK (but with many warnings In file included from i:/boost-trunk/libs/test/src/test_tools.cpp:16: In file included from i:/boost-trunk\boost/test/impl/test_tools.ipp:23: i:/boost-trunk\boost/test/tree/test_unit.hpp:164:7: warning: 'dllexport' attribute only applies to variables and functions class BOOST_TEST_DECL master_test_suite_t : public test_suite { ^ i:/boost-trunk\boost/test/detail/config.hpp:106:29: note: expanded from macro 'BOOST_TEST_DECL' # define BOOST_TEST_DECL BOOST_SYMBOL_EXPORT ^ i:/boost-trunk\boost/config/platform/win32.hpp:30:42: note: expanded from macro 'BOOST_SYMBOL_EXPORT' # define BOOST_SYMBOL_EXPORT __declspec(dllexport) ^ <built-in>:150:38: note: expanded from macro '__declspec' #define __declspec(a) __attribute__((a)) but the link stage to build the library fails with many Cannot export boost::test_tools::tt_detail::check_impl(boost::test_tools::predicate_result const&, boost::unit_test::lazy_ostream const&, boost::unit_test::basic_cstring<char const>, unsigned int, boost::test_tools::tt_detail::tool_level, boost::test_tools::tt_detail::check_type, unsigned int, ...): symbol not found (full output window message attached). I have also failed to build using bjam (missing <cstddef> so missing the right stdlib?) and I wonder if both these are connected with lack of a suitable boost/config.hpp which caters for Clang *on this configuration*. Where would/should __clang__ be defined in order to select this compiler? My user-config.jam contains (in an attempt to cure ) # Clang compiler using mingw32 dw2 Ruben's release. using clang : 3.1 : "C:/mingw32-dw2/bin/clang++.exe" : <cxxflags>-v <cxxflags>-std="c++11" <cflags>-nostdinc # Do not include MSVC C headers. #<cflag>-c-include-dirs=??? <cxxflags>-nostdinc++ # Do not include MSVC C++ headers. # Use mingw-dw2 for includes <cxxflags>-cxx-include-arch="C:/mingw32-dw2/i686-w64-mingw32" <cxxflags>-cxx-include-32bit-dir=32 <cxxflags>-cxx-include-root="C:/mingw32-dw2/include/c++/4.6.3" <cxxflags>-stdlib=libc++ #<compileflags>-std=gnu++11 #<define>BOOST_STDLIB_CONFIG="i:/boost-trunk/boost/config/stdlib/libstdcpp3.hpp" <linkflags>-stdlib=libc++ ; Any views on my woes? Paul --- Paul A. Bristow, Prizet Farmhouse, Kendal LA8 8AB UK +44 1539 561830 07714330204 pbristow(a)hetp.u-net.com

2 7

Re: [boost] Graph Algorithm Advice
by J Birch 19 Nov '12

19 Nov '12

Sorry, I had pasted it in the client. it is attached now. Thanks! Birch On Nov 19, 2012, at 5:03 PM, Jeremiah Willcock <jewillco(a)osl.iu.edu> wrote: > (off list) > > I don't see your attachment on there. > > -- Jeremiah Willcock > > On Mon, 19 Nov 2012, J Birch wrote: > >> This isn't a Boost Graph problem per se, just a graph algorithm question. Let's say I have the following graph: >> >> >> >> I want to represent this graph in a certain way. In the small legend box above, there are 2 versions of a correct answer. I want to print out each vertex followed by it's immediate adjacent vertex. If it has more than one, then each path from that point should be wrapped in a '()'. Currently, I am using the Ruby RGL package and have tried both a directed and an undirected graph using a depth first search traversal, and have yet to hit on an algorithm that takes every edge case. >> >> Anyone out there have some pointers on how to do this? >> >> Thanks! >> Birch >> >> _______________________________________________ >> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost >>

1 0

Re: [boost] Countertree
by Leo Goodstadt 19 Nov '12

19 Nov '12

On 17 November 2012 19:44, Francisco Jos? Tapia <fjtapia(a)gmail.com> wrote: > 2012/11/16 Vadim Stadnik <vadimstdk(a)gmail.com> > > On Sat, Nov 17, 2012 at 4:25 AM, Leo Goodstadt <leo.goodstadt(a)llew.org.uk wrote: > > > > > > Project location ( zip file with code and documentation) : > > > > https://dl.dropbox.com/u/8437476/works/countertree_code_doc.zip > > > > > > > > Quick view of documentation with code download : > > > > https://dl.dropbox.com/u/8437476/works/countertree/index.html > > > > > > > > This library is an implementation of a binary red-black counter tree. > > > > This tree have an additional counter in each leaf. This permit the access to > > > > the elements by the position, like in a vector. It is a random access > > > > container with random access iterators > > > > > > > > > Among other things, countertree would allow the boost Statistical > > > Accumulators Library to support rolling median and rolling quantiles > > > efficiently. This would be very useful. > > > > > > I had previously had to hack the tree class in the Policy-Based Data > > > Structures part of the gnu c++ library. > > > (see http://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/tree_order_statistics_nod… ) > > As far as I know, this support is not yet available in countertree project, > > although this is possible in principle. > > About the compatibility with the boost Statistical Accumulators Library, I > didn't thought about it, but it would be useful. I will try to implement, > but I need to examine in deep for to provide you a well founded opinion. > 1) For counter trees, the median is just the N/2-th value in the counter tree, where K is the size of the tree. The Nth-percentile can be calculated similarly. A rolling median with a fixed size window requires there to be an insertion and deletion for each (most!) new value, so I am guessing it should run in O(N log K) time, where N = size of data, K = size of window 2) The alternative is to use two multisets corresponding to values smaller and larger than the current median (or the Nth percentile). This should also have O(N log K) complexity. 3) With the special case where the data consists of integers in a fixed range, we can use a segment tree where the complexity would be O(N log R) where R = the size of the fixed range of data values. This can result in considerable speed ups. The best bit about counter trees are their flexibility, we can calculate multiple Nth-percentiles at the same time (e.g. all the quartiles). As to the actual rather than theoretical speed, I think we need bench-marking.... Leo

2 1