April 2004 - Boost-users - lists.boost.org

Selection scheme for boost libraries
by Uli Tuerk 19 Apr '04

19 Apr '04

Hi all! I have a very basic problem with the boost libraries: I used bjam to build all boost libraries for Linux (using "gcc" as toolset and the call 'bjam "-sTOOLS=gcc" install'). All went fine, however I wanted to do the regression tests for boost_regex using the gcc.mak Makefile in <boost_main>/libs/regex/test/regress which uses "-lboost_regex" for linking. The linker reported an error because there is no "libboost_regex.so" in my lib folder unless I created a link with this name to one of the various versions of libboost_regex. A comment for these Makefiles would be helpful for the case when having build all with bjam. When compiling the regex part alone, the library is created in build/gcc with the correct name (libboost_regex.so). I'm new to C++, so maybe I haven't got the point with the sophisticated naming scheme of the libraries, but: for the library "libboost_regex-gcc-mt-d-1_31.so" e.g. are the tool specifier ("gcc") and the version ("1_31") really necessary for a platform using version numbering of the libraries anyway in the ".so" extension? OK, there are still the categories with/without debug and with/without threading. How would you guys select the type of library? In the Makefile? Thanks in advance! Uli ************************************************************************ Uli Türk

2 1

BGL assertion
by Val Samko 18 Apr '04

18 Apr '04

Something triggers the assert line in push_relabel_max_flow, assert(algo.is_flow()); is that internal push_relabel_max_flow problem, or this means I supply invalid data to this function? Val Samko http://val.digiways.com

1 0

Problems compiling boost.python with Boost 1.31.0
by Markus Sch�pflin 17 Apr '04

17 Apr '04

I'm trying to compile boost.python from Boost 1.31.0 on Tru64 5.1 with Compaq C++ V6.5-033. Python has been built like this: ./configure --with-cxx=cxx --prefix=/opt --exec-prefix=/opt/exec/OSF1-V5.1-a lpha --without-gcc The command line used to build boost.python looks like this: bjam -sTOOLS=tru64cxx65 -sPYTHON_ROOT="/opt" -sPYTHON_VERSION="2.3" stage When build boost.python I get the following error message: tru64cxx65-C++-action bin/boost/libs/python/build/libboost_python.so/tru64cxx65/debug/numeric.o cxx: Error: /net/camos/opt/include/python2.3/Python.h, line 8: #5 could not open source file "pyconfig.h" #include "pyconfig.h" ---------------------^ It looks like the exec prefix is not taken into account because the file is located here: /opt/exec/OSF1-V5.1-alpha/include/python2.3/pyconfig.h Is this not supported by the current boost build process or is there any option I missed? Or is there anything else I'm doing wrong? Thanks, Markus

3 3

broken link in boost.org FAQ
by Mike Morearty 17 Apr '04

17 Apr '04

Minor problem: on the FAQ at http://www.boost.org/more/faq.htm , the very first question is, "How do I download the libraries?" and the answer is, "See Download instructions." But "Download" is a broken link. Points to http://www.boost.org/libs/libraries.htm#Download ; although http://www.boost.org/libs/libraries.htm does exist, it does not contain an anchor called "Download". The FAQ should probably be changed to point to http://www.boost.org/more/getting_started.html , or maybe to http://sourceforge.net/project/showfiles.php?group_id=7586 - Mike Morearty

1 0

[MPL] Runtime Interaction
by Debian User 16 Apr '04

16 Apr '04

[Sent via gmane but it never appeared] Hi gang, Suppose I have a number of types that each have a "value" enum, much like MPL's constant classes. Now suppose that I want to construct a runtime-indexable array of those values, according to the position of each type in an MPL vector. For example: int values[3]; typedef boost::mpl::vector_c<int,10, 11, 12> values; typedef some_magic_metafunction<values>::type runtime_function; // values[0] = 10, values[1] = 11, values[2] = 12 runtime_function::fill_array(values); The idea is that some_magic_metafunction<values> would generate a type kind of like this: template<typename TypeValueList, typename Index> struct magic { static void fill_array(int *v) { // Non-MPL type vector indexing -- could use an // MPL iterator here v[Index::value] = TypeValueList<Index::value>::value; if (Index::value < distance<begin<TypeValueList>::type, end<TypeValueList>::type>::type::value) { magic<TypeValueList, Index::next>::fill_array(v); } }; }; Get it? The problem is that I don't know how to make MPL transition from strictly compile-time to something in-between. It can statically know the size of the value array and the values themselves but it has to fill the array "values" at runtime so that the values may be indexed by runtime computations. Any guidance? Thanks! -Dave

2 1

problems using VC 7.1 and boost regex library
by Keith Smith 15 Apr '04

15 Apr '04

I have been trying to compile the following code with VC 7.1 ( which compiled fine with VC 6 ) with no luck. std::string m_rBuffer; boost::regex m_rStartExp; string::const_iterator rRawDataBegin = m_rBuffer.begin(); string::const_iterator rRawDataEnd = m_rBuffer.end(); boost::cmatch rStartMessage; unsigned int flags = boost::match_default; while ( boost::regex_search(rRawDataBegin, rRawDataEnd, rStartMessage, m_rStartExp, flags) ) Included is the compiler error report. Please throw me a bone.

2 1

BGL linking on Win32
by maxx＠dgap.mipt.ru 15 Apr '04

15 Apr '04

Hi! whenever I use libbgl-viz.lib to read graphviz files, I get a linker errors like this: msvcprt.lib(MSVCP71.dll) : error LNK2005: "public: __thiscall std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> >::basic_string<char,struct std::char_traits<char>,class std::allocator<char> >(class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &)" (??0?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@QAE@ABV01@@Z) already defined in Graphviz.obj It was Ok to add "/NODEFAULTLIB:msvcrt" to the compiler options, but now I need that library! So if I don't add this option, I get errors like the one above, if I do add it, I get errors "unresolved symbols"... What is probably(or definitely) wrong? Thanks in advance, -- Maxim

2 1

RE: [Boost-users] Changing the heap allocator
by Jim.Hyslop 14 Apr '04

14 Apr '04

David Barrett-Lennard wrote: > Yes HeapSetInformation() will do it once you have the heap handle. > However, the CRT creates its own heap with a call to HeapCreate(), and > stores the heap handle in a global variable called _crtheap > that is not > exported by the library. Right, of course - I thought that GetProcessHeap() would return the correct heap, but it does not. -- Jim Hyslop Senior Software Designer Leitch Technology International Inc. (http://www.leitch.com) Columnist, C/C++ Users Journal (http://www.cuj.com/experts)

1 0

RE: [Boost-users] Changing the heap allocator
by Jim.Hyslop 14 Apr '04

14 Apr '04

David Barrett-Lennard wrote: > Does anyone know how to enable the LFH (low fragmentation > heap) for the > heap created by the VC CRT? Have you tried HeapSetInformation()? -- Jim Hyslop Senior Software Designer Leitch Technology International Inc. (http://www.leitch.com) Columnist, C/C++ Users Journal (http://www.cuj.com/experts)

2 1

Changing the heap allocator
by David Barrett-Lennard 13 Apr '04

13 Apr '04

Hi, I'm working on a project where I've replaced Microsoft's (terrible) heap allocator with the Doug Lea allocator. There seems to be some cases where the Windows XP allocator is 500 times slower than it should be. This made my program take minutes rather than seconds to run. To try to understand what kind of allocation patterns cause the behaviour I wrote a small console application (see below). After playing around with different allocation patterns for half an hour I found a case where the pathological behaviour appeared. I'm none the wiser about the problem. For example, moving the malloc(301) statement before the 7 x malloc(1000) makes the problem go away! However, given the ease with which I rediscovered the problem, I wonder why there aren't more complaints about Microsoft's heap allocator. With some allocation patterns I have seen the allocator fall below 1kHz on a Pentium IV 2GHz machine! I've created a library that exports alternative versions of operator new and operator delete, by using a def file as follows EXPORTS ??2@YAPAXI@Z ??3@YAXPAX@Z I've had to do some hacky things to make it work with the STL (because of msvcp60.dll) , but that's another story! Now my question : I want to use boost, but need to work out how to make it use a different allocator. Will I need to rebuild the boost dlls (such as boost_python.dll), against my static library? Regards, David Barrett-Lennard /////////////////////// heap allocation test /////////////////////////// #include <iostream> #include <time.h> int main(int argc, char* argv[]) { for (int count=1 ; count <= 5 ; ++count) { int n1 = 0, n2 = 0; clock_t start = clock(); { const int NUM = 10000; void* L[NUM*7]; for (int i=0 ; i < NUM ; ++i) { for (int j=0 ; j < 7 ; ++j) L[n1++] = malloc(1000); void* leak = malloc(301); ++n2; } for (int k=n1-1 ; k >= 0 ; --k) free(L[k]); } clock_t elapsed = clock()-start; if (elapsed) { std::cout << "Time = " << elapsed << " ms" << " Rate = " << 1000 * (n1 + n2) / elapsed << " Hz" << std::endl; } } return 0; }

3 4