- Boost-users - lists.preview.boost.org

two boost.threads taking much longer than one
by Andrew Chapman 11 Oct '04

11 Oct '04

Hi everyone, I've got an operation which takes 14 seconds on a single thread, but is taking up to 45 seconds when running via two boost.threads on a dual proc machine (2x 3ghz Xeons, Fedora Core 1). The thread functor objects are being passed a lot of references (about 20) to some very large std::vectors. One thread always reads/writes from the start of the vectors to their midpoint, and the second thread from (midpoint+1) to the end of the vectors. The only thing I can think of is that the threads are taking a long time to fork and get running, perhaps moving a lot of stack data around or something? When I look at the output of 'top' with threads displayed, I see the two threads appear, but they only use about 50-60% CPU utilization each. The system monitor shows mostly red (Kernel) as opposed to blue (User). Any ideas? -- Andrew Chapman Senior Technical Director - Framestore CFC

3 3

Re: Re: [Boost-users] function_traits inheritance
by Victor Morilla 11 Oct '04

11 Oct '04

Peter Dimov wrote: >Victor Morilla wrote: >> >> template <typename FnType, FnType* fn> >> struct Translator<FnType, fn, 0> : >> function_traits<FnType> >> { >> typedef result_type type; > typedef typename function_traits<FnType>::result_type type; >> }; The question is, why result_type type definition is not inherited from function_traits ? ______________________________________________ Renovamos el Correo Yahoo!: ¡100 MB GRATIS! Nuevos servicios, más seguridad http://correo.yahoo.es

2 1

function_traits inheritance
by Victor Morilla 11 Oct '04

11 Oct '04

Guess someone why this code fails ? (g++3.3) ---- begin ----------------------- #include <boost/type_traits.hpp> #include <iostream> using namespace std; using namespace boost; int foo() { return 10; } template <typename FnType, FnType* fn, int Arity = function_traits<FnType>::arity> struct Translator; template <typename FnType, FnType* fn> struct Translator<FnType, fn, 0> : function_traits<FnType> { typedef result_type type; }; int main(int argc, char** argv) { Translator<int (), &foo>::type x; } ------------------- end code ------------------ error: ISO C++ forbids declaration of `result_type' with no type --------------------- end error ------------------- And another question: is there a portable equivalent to: Translator<typeof(foo), &foo> Thanks ______________________________________________ Renovamos el Correo Yahoo!: ¡100 MB GRATIS! Nuevos servicios, más seguridad http://correo.yahoo.es

3 2

Problems in compilation with Boost
by Debayan Chaudhuri 11 Oct '04

11 Oct '04

Hi, I am evaluating Boost for use in our project. The platform OS is FreeBSD 4.9; which comes with gcc-2.95.3. I downloaded boost_1.30.0 and compiled the thread library on the FreeBSD box. When I try to compile a small test program using Boost threads; I encounter hundreds of compiler errors; mostly in the Boost header files. Is there a workaround to this ? Which version of Boost works with gcc-2.95.3 and what are the compiler options that will make it compile properly? regards, Debayan.

2 2

RE: [Boost-users] Problems with compiling Filesystem library on Solaris
by Barak Simon 11 Oct '04

11 Oct '04

By now I had time to check your suggestion, but I'm afraid I still get the same error when building boost libs: "/db/pub/eq/build/src/boost_1_31_0/boost/iterator/iterator_facade.hpp", line 622: Error: Could not find a match for boost::iterator_core_access::equal<boost::iterator_core_access::Facade1, boost::iterator_core_access::Facade2>(const boost::filesystem::directory_iterator, const boost::filesystem::directory_iterator, boost::is_convertible<boost::filesystem::directory_iterator, boost::filesystem::directory_iterator>). "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../src/operations_posix_windows.cpp", line 176: Where: While instantiating "boost::operator==<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int, boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>(const boost::iterator_facade<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>&, const boost::iterator_facade<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>&)". "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../src/operations_posix_windows.cpp", line 176: Where: Instantiated from non-template code. Are you sure that is the only line I need to comment out? Thanks, Barak ________________________ Barak Simon GED IT Core Platform 190 George St. Sydney, 2000 Phone +61 2 925 85070 Fax +61 2 925 95050 barak.simon+external@db. com To: boost-users(a)lists.boost.org Sent by: cc: boost-users-bounces@list Subject: RE: [Boost-users] Problems with compiling Filesystem library on Solaris s.boost.org 11/10/2004 11:45 Please respond to boost-users Thanks for the swift reply! In the interim I commented out the 2-3 problematic methods (which I am not using anyway) in the filesystem lib, but I like your solution better. A question on that though: On your solution, what exactly am I commenting out, that is, can the workaround break some other boost library or a significant assumption? Thanks, Barak Tomas.Puverle@morganstan ley.com To: boost-users(a)lists.boost.org Sent by: cc: boost-users-bounces@list Subject: RE: [Boost-users] Problems with compiling Filesystem library on Solaris s.boost.org 08/10/2004 18:47 Please respond to boost-users > Hi, > I get the following error from the Filesystem library when > building boost 1_31 on Solaris with sunpro compiler: > > "/db/pub/eq/build/src/boost_1_31_0/boost/iterator/iterator_fac > ade.hpp", line 622: Error: Could not find a match for > boost::iterator_core_access::equal<boost::iterator_core_access > ::Facade1, boost::iterator_core_access::Facade2>(const > boost::filesystem::directory_iterator, const > boost::filesystem::directory_iterator, > boost::is_convertible<boost::filesystem::directory_iterator, > boost::filesystem::directory_iterator>). > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: While instantiating > "boost::operator==<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int, > boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>(const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&, const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&)". > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: Instantiated from non-template code. > > > > It looks suspiciously similar to bug "[ 976241 ] Compiler > error for tokenizer on Solaris" from 2004-06-20, with the > difference that I am using version 5.3 of sunpro rather than 5.5. I have submitted a similar problem and the proposed fix recently. I am just waiting on our developers to confirm that it works before submitting it to boost, but it seems to fix the tokenizer problem, too. For a quick workaround (before the patch is applied) do this: At the top of boost/type_traits/detail/bool_trait_def.hpp you will see something like this: #if defined(__SUNPRO_CC) # define BOOST_TT_AUX_BOOL_TRAIT_VALUE_DECL(C) \ typedef mpl::bool_< C > type; \ enum { value = type::value }; \ /**/ # define BOOST_TT_AUX_BOOL_C_BASE(C) //1 Comment out the line marked with //1. That seems to have fixed my problems with both 5.3 and 5.5. HTH, Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -- This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -- This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

1 0

Function type transformation and manipulation traits?
by Aaron W. LaFramboise 11 Oct '04

11 Oct '04

Are there any Boost support for manipulating function types? Operations I have in mind include: setting return type, setting argument type, changing number of arguments, et cetera. If not, is there some generally accepted way to accomplish this, and is any work being done to add this feature to Boost? Aaron W. LaFramboise

2 3

RE: [Boost-users] Problems with compiling Filesystem library on Solaris
by Barak Simon 11 Oct '04

11 Oct '04

Thanks for the swift reply! In the interim I commented out the 2-3 problematic methods (which I am not using anyway) in the filesystem lib, but I like your solution better. A question on that though: On your solution, what exactly am I commenting out, that is, can the workaround break some other boost library or a significant assumption? Thanks, Barak Tomas.Puverle@morganstan ley.com To: boost-users(a)lists.boost.org Sent by: cc: boost-users-bounces@list Subject: RE: [Boost-users] Problems with compiling Filesystem library on Solaris s.boost.org 08/10/2004 18:47 Please respond to boost-users > Hi, > I get the following error from the Filesystem library when > building boost 1_31 on Solaris with sunpro compiler: > > "/db/pub/eq/build/src/boost_1_31_0/boost/iterator/iterator_fac > ade.hpp", line 622: Error: Could not find a match for > boost::iterator_core_access::equal<boost::iterator_core_access > ::Facade1, boost::iterator_core_access::Facade2>(const > boost::filesystem::directory_iterator, const > boost::filesystem::directory_iterator, > boost::is_convertible<boost::filesystem::directory_iterator, > boost::filesystem::directory_iterator>). > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: While instantiating > "boost::operator==<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int, > boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>(const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&, const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&)". > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: Instantiated from non-template code. > > > > It looks suspiciously similar to bug "[ 976241 ] Compiler > error for tokenizer on Solaris" from 2004-06-20, with the > difference that I am using version 5.3 of sunpro rather than 5.5. I have submitted a similar problem and the proposed fix recently. I am just waiting on our developers to confirm that it works before submitting it to boost, but it seems to fix the tokenizer problem, too. For a quick workaround (before the patch is applied) do this: At the top of boost/type_traits/detail/bool_trait_def.hpp you will see something like this: #if defined(__SUNPRO_CC) # define BOOST_TT_AUX_BOOL_TRAIT_VALUE_DECL(C) \ typedef mpl::bool_< C > type; \ enum { value = type::value }; \ /**/ # define BOOST_TT_AUX_BOOL_C_BASE(C) //1 Comment out the line marked with //1. That seems to have fixed my problems with both 5.3 and 5.5. HTH, Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -- This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

1 0

Help ... BOOST::thread 's certain constuctor methods fails with STLPort
by Raj 09 Oct '04

09 Oct '04

Hi I have build boost libraries for a) regular gcc b) gcc-STLPort the following code works perfectly for "gcc" and throws up the following errors for "gcc-STLPort". // this is my function object that // takes in input parameters as part of its constructor. func_obj instance(data,m_regMap); boost::shared_ptr<thread> pThread(new boost::thread(instance)); ERROR: mystuff.cpp:101: undefined reference to `boost::thread::thread[in-charge] (boost::function0<void, _STL::allocator<boost::funct ion_base> > const&)' If I assume that the boost runtime library has not been linked, that is not the case.(ldd tells me that i have) whereever I use 'xtime','lock' or 'condition' it doesn't complain. WHAT IS STRANGE IS THE FOLLOWING COMPILES boost::thread thrd; boost::thread thrd(func_obj); // it is the actual fo not an instance. Any help would be greatly appreciated. This happens on Linux/with STLPort 4.6.2/gcc 3.2.3

1 0

How to compile BOOST on Linux with NPTL threads and STLPort
by Raj 09 Oct '04

09 Oct '04

Hi, I am trying to use BOOST thread and regex libraries on Linux. I am able to compile them with sTOOLSS=gcc-stlport option. But, I would also like to specify that I would like to use the NPTL thread lib instead of pthread. How can I do it ? Any help would be greatly appreciated. Thanks in Advance Raja

2 3

building graph library
by kyle huang 09 Oct '04

09 Oct '04

i just started looking at the boost libraries and thus i'm a total beginner. i'm trying to build the graph library in particular, and i haven't had any luck at all. first, is it a buildable library, or is it just a collection of templates? if so (collection of templates), how do i use it? i'm using windows xp, and building for Visual Studio 6 thanks!

3 2

BLL: Confusing sig example code in documentation
by Markus Larsson 09 Oct '04

09 Oct '04

from the BLL/bind documentaion on the cvs: struct A { // the return type equals the third argument type: template<class T1, T2, T3> T3 operator()(const T1& t1, const T2& t2, const T3& t3); template <class Args> class sig { // get the third argument type (4th element) typedef typename boost::tuples::element<3, Args>::type T3; public: typedef typename boost::remove_cv<T3>::type type; } }; To begin with this piece of code has 2 syntax errors. template<class T1, T2, T3> // should have class before all parameters struct sig { ... } // missing semicolon If a first time user copy/pastes the example, fixes the syntax errors, defines operator() and tries to call the function with: int i = 7; bind( A(), 1, 2, _1) (i); she gets a hard to parse error because operator() wasn't declared const. I see two possible solutions: 1) Add const to the declaration 2) mention the non constness and show how to call the function object using var(func), with a short note on why that is needed.

1 0

shared_ptr interactions
by eric lindvall 08 Oct '04

08 Oct '04

I have created a simple template to make instantiation of shared_ptrs a little prettier. Here is an abridged version of the template: template <class T> class ManagedPointer : public boost::enable_shared_from_this<T> { public: typedef boost::shared_ptr<T> shared_ptr; public: static shared_ptr create () { return (shared_ptr (new T ())); } template <class A1> static shared_ptr create (A1 a1) { return (shared_ptr (new T (a1))); } }; This allows for me to create a class: class TestClass : public ManagedPointer<TestClass> { public: TestClass(); }; and create an instance by: TestClass::shared_ptr test = TestClass::create(); I have noticed in certain cases I get very strange behaviour. Most recently I've seen things hanging in the shared_ptr destructor. Changing the code to: TestClass::shared_ptr test (new TestClass()); does "solve" the problem. Is there something subtle in how the templates and subclassing are interacting that I am not understanding? thanks. e.

2 2

boost implementation
by Gast 128 08 Oct '04

08 Oct '04

Dear all, Browsing through source files of boost, I see that some libraries reimplement common idioms found in other libraries. For example the tuple library reimplements its own compile time if and it's own typelist. I guess it would be better if high level boost libraries(e.g. regex, threads, to some extend also the tuples) would use the low level boost libraries (e.g. preprocessor, type_traits, mpl). Besides bugfixes and platform/compiler workarounds, it has also a documentary advantage. If you are familiar with mpl for example, one knows how to interpret the construct with if_. Also one can better interact with libraries if they are using the same base constructs. There almost no disadvantages, because all the low level boost libraries are template based. But probably this is a matter of time, because at first look at the variant library, I see that it uses the mpl. And people are talking about the fusion library for the tuples. Are there boost developers who can comment on this? Wkr, me.

2 1

RE: [Boost-users] Problems with compiling Filesystem library on Solaris
by Puverle, Tomas (IT) 08 Oct '04

08 Oct '04

> Hi, > I get the following error from the Filesystem library when > building boost 1_31 on Solaris with sunpro compiler: > > "/db/pub/eq/build/src/boost_1_31_0/boost/iterator/iterator_fac > ade.hpp", line 622: Error: Could not find a match for > boost::iterator_core_access::equal<boost::iterator_core_access > ::Facade1, boost::iterator_core_access::Facade2>(const > boost::filesystem::directory_iterator, const > boost::filesystem::directory_iterator, > boost::is_convertible<boost::filesystem::directory_iterator, > boost::filesystem::directory_iterator>). > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: While instantiating > "boost::operator==<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int, > boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>(const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&, const > boost::iterator_facade<boost::filesystem::directory_iterator, > boost::filesystem::path, boost::single_pass_traversal_tag, > boost::filesystem::path&, int>&)". > > "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../sr c/operations_posix_windows.cpp", line 176: Where: Instantiated from non-template code. > > > > It looks suspiciously similar to bug "[ 976241 ] Compiler > error for tokenizer on Solaris" from 2004-06-20, with the > difference that I am using version 5.3 of sunpro rather than 5.5. I have submitted a similar problem and the proposed fix recently. I am just waiting on our developers to confirm that it works before submitting it to boost, but it seems to fix the tokenizer problem, too. For a quick workaround (before the patch is applied) do this: At the top of boost/type_traits/detail/bool_trait_def.hpp you will see something like this: #if defined(__SUNPRO_CC) # define BOOST_TT_AUX_BOOL_TRAIT_VALUE_DECL(C) \ typedef mpl::bool_< C > type; \ enum { value = type::value }; \ /**/ # define BOOST_TT_AUX_BOOL_C_BASE(C) //1 Comment out the line marked with //1. That seems to have fixed my problems with both 5.3 and 5.5. HTH, Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

1 0

Problems with compiling Filesystem library on Solaris
by Barak Simon 08 Oct '04

08 Oct '04

Hi, I get the following error from the Filesystem library when building boost 1_31 on Solaris with sunpro compiler: "/db/pub/eq/build/src/boost_1_31_0/boost/iterator/iterator_facade.hpp", line 622: Error: Could not find a match for boost::iterator_core_access::equal<boost::iterator_core_access::Facade1, boost::iterator_core_access::Facade2>(const boost::filesystem::directory_iterator, const boost::filesystem::directory_iterator, boost::is_convertible<boost::filesystem::directory_iterator, boost::filesystem::directory_iterator>). "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../src/operations_posix_windows.cpp", line 176: Where: While instantiating "boost::operator==<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int, boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>(const boost::iterator_facade<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>&, const boost::iterator_facade<boost::filesystem::directory_iterator, boost::filesystem::path, boost::single_pass_traversal_tag, boost::filesystem::path&, int>&)". "/db/pub/eq/build/src/boost_1_31_0/libs/filesystem/build/../src/operations_posix_windows.cpp", line 176: Where: Instantiated from non-template code. It looks suspiciously similar to bug "[ 976241 ] Compiler error for tokenizer on Solaris" from 2004-06-20, with the difference that I am using version 5.3 of sunpro rather than 5.5. My guess is there must be some flag/macro/workaround I am missing (as this specific library is supposed to be portable to many OSs and compilers) , but have no idea what that may be. Any suggestions would be a blessing. Thanks, Barak ________________________ Barak Simon GED IT Core Platform Deutsche Bank 190 George St. Sydney, 2000 Phone +61 2 925 85070 Fax +61 2 925 95050 -- This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

1 0

Typelists
by Puverle, Tomas (IT) 07 Oct '04

07 Oct '04

What is the status of typelist support in boost? I read through a backlog of email in the archives but the last ones are from ~2002. Grepping through the source of 1.31.0 for (typelist | type_list) reveals some in the detail namespace for python, but that's all. Thanks, Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

2 1

uBLAS lapack bindings
by Daniel L Elliott 07 Oct '04

07 Oct '04

Hello, I have a question about using uBLAS/LAPACK bindings on AIX 5.1 and VA6. I am doing something like the following: matrix<double> matrixTwo(200,200); identity_matrix<double> identMatrix = identity_matrix<double>(200); int inverseInfo = gesv(matrixTwo,identMatrix); I am getting the following errors: line 102.36: 1540-0130 (S) "matrix_traits<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::a\ llocator<double> > > >::size1" is not declared "/npd/delliott/boost/numeric/bindings/traits/matrix_traits.hpp", line 102.7: 1540-0700 (I) The previous message was produced while processing "boost::numeric::bindings::traits::matrix_size1<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > > >(matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > > &)". "/npd/delliott/boost/numeric/bindings/lapack/gesv.hpp", line 135.40: 1540-0700 (I) The previous message was produced while processing "boost::numeric::bindings::lapack::gesv<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > >,boost::numeric::ublas::identity_matrix<double> >(matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double, std::allocator<double> > > &, identity_matrix<double> &)". "test.cpp", line 120.21: 1540-0700 (I) The previous message was produced while processing "main()". "/npd/delliott/boost/numeric/bindings/lapack/gesv.hpp", line 99.9: 1540-0038 (S) The name "typename traits::matrix_traits<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > >>::matrix_structure" is not a type. "/npd/delliott/boost/numeric/bindings/lapack/gesv.hpp", line 95.9: 1540-0700 (I) The previous message was produced while processing "boost::numeric::bindings::lapack::gesv<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > >,boost::numeric::ublas::identity_matrix<double>,boost::numeric::bindings::traits::detail::array<int> >(matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > > &, array<int> &, identity_matrix<do...". "/npd/delliott/boost/numeric/bindings/lapack/gesv.hpp", line 137.16: 1540-0700 (I) The previous message was produced while processing "boost::numeric::bindings::lapack::gesv<boost::numeric::ublas::matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double,std::allocator<double> > >,boost::numeric::ublas::identity_matrix<double> >(matrix<double,boost::numeric::ublas::row_major,boost::numeric::ublas::unbounded_array<double\ ,std::allocator<double> > > &, identity_matrix<double> &)". As I understand it, I can simply pass the uBLAS matrix types and the traits/lapack .hpp files will take care of the rest. Any light that could be shed upon this subject would be of great assistance. By the way, I would have happily send this to uBLAS-dev, but have not been admitted yet :( Thank you, Dan

2 1

FW: Compilation problem with boost on SunCC (Forte 6.2 & 8)
by Puverle, Tomas (IT) 06 Oct '04

06 Oct '04

A repost, since no answer to my previous mail. > -----Original Message----- > From: Puverle, Tomas (IT) > Sent: 01 October 2004 15:05 > To: 'boost(a)lists.boost.org' > Subject: Compilation problem with boost on SunCC (Forte 6.2 & 8) > > Please find attached a small program and a the resulting > compilation errors on SunCC. > This occurs in iterator_facade.hpp, operator != (), calling > to interator_core_access::equal. I have tracked this down to > the following problem: > > f(mpl::true_); > f(mpl::false_); > > f(is_convertible<>()); //will not work > > is_convertible<> derives from mpl::bool<>, however it seems > that a bug in the compiler prevents it from converting to > mpl::true_/mpl::false_ in the above scenario (and in the > program attached) > > A workaround for this is to change it to the following: > > f(mpl::true_); > f(mpl::false_); > > f(mpl::bool_<is_convertible<>::value>()); > > Presumably this will be present in other places but this is > the only one our developers have hit so far. > > Thanks, > > Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

3 3

Re: Problem compiling Windows C++ code wit regex
by arno.schaefer＠sqs.de 06 Oct '04

06 Oct '04

> Yes, that occurs if you have a > > using namespace std; > > statement in your code, and you are using Microsoft Visual C++ 6 (it's > actually a compiler bug). > > Removing (or possibly moving until after all your includes, or just > re-ordering your includes) the using statement will fix the issue. > Thanks very much for that hint John, I found two places in our header which are using the "using namespace std;". After I have switched that to the explicit "using" statements I could also switch back to your default "user.hpp". Perhaps I can ask one more question about this strange behaviour, just because I want to understand why: I try to use my own string class, which is inherited from std::string. Defining "VC6_STUPID_BEHAVIOUR" in the example code below compile and run the code successfuly, using the expanded name leads to following compiler error: error C2614: 'MyCString' : illegal member initialization: 'string' is not a base or member You know WHY??? Also because of the same compiler error in VC6 or is there something else wrong. On Solaris and HP-UX it works fine in every combination. // vvvvvvvvvvvvvvv snipp vvvvvvvvvvvvvvvv #include <string> #include <iostream> #ifdef VC6_STUPID_BEHAVIOUR using std::string; # define std_string string #else # define std_string std::string #endif class MyCString : public std_string { public: MyCString(const char *s_); : std_string(s_) { std::cout << "\"" << *this << "\"" << std::endl; }; }; int main (int argc, char *argv[]) { MyCString s(argv[0]); return 0; } // ^^^^^^^^^^^^ snipp ^^^^^^^^^^^^

2 1

boost install with gentoo issue?
by dave＠superelite.net 06 Oct '04

06 Oct '04

Well, more likely my mistake, ive been up awhile.. I installed the boost libs onto my gentoo machine with emerge. I then added -lboost_regex to my linker flags..thats it. When I run my app, it hangs, I think when waiting on pulling data of off a UDP socket. Though I cant tell, I'm about to crash and its been one of those days. But I do know everything is back to normal when I take -lboost_regex out of the flags again. I know I'm awfully vague due to lazy/tiredness, but does this ring a bell with anyone? I've installed boost the traditional way (bjam) on other linux distros before without any issue, but ive never tried emerge before. I dont know if this has something to do with it, or i am somehow missing something really obvious.. Thanks for bearing with me, Dave

4 4

uBLAS LAPACK bindings
by Daniel L Elliott 06 Oct '04

06 Oct '04

Hello, Are the uBLAS LAPACK bindings for general matrices or for any type of matrix? Do they have to use a particular type of storage? This is how it appears in the code. Would anyone like to talk about how other types of matrices might be used? Thank you, Dan

2 1

RE: [Boost-users] uBlas + ATLAS...
by Lyon, Richard 06 Oct '04

06 Oct '04

Hmm, Atlas requires cygwin as minimum on WinXP. The install won't work with msys from mingw. Regards ... -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of DrJekyll Sent: Wednesday, 6 October 2004 4:41 AM To: boost-users(a)lists.boost.org Subject: [Boost-users] uBlas + ATLAS... dear all, I have a question about uBlas and ATLAS, and somewhat off topic. In fact , i try to install ATLAS under WIN XP (with out cygwin) with out success. My question is have any one succeed in installing this library and whether uBlas can be used in the place of Atlas. Thank for your attention. JK _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

uBlas + ATLAS...
by DrJekyll 05 Oct '04

05 Oct '04

dear all, I have a question about uBlas and ATLAS, and somewhat off topic. In fact , i try to install ATLAS under WIN XP (with out cygwin) with out success. My question is have any one succeed in installing this library and whether uBlas can be used in the place of Atlas. Thank for your attention. JK

1 0

Compilation problem with boost on SunCC (Forte 6.2 & 8)
by Puverle, Tomas (IT) 05 Oct '04

05 Oct '04

A repost, since no answer to my previous mail. > > Please find attached a small program and a the resulting > compilation > > errors on SunCC. > > This occurs in iterator_facade.hpp, operator != (), calling to > > interator_core_access::equal. I have tracked this down to the > > following problem: > > > > f(mpl::true_); > > f(mpl::false_); > > > > f(is_convertible<>()); //will not work > > > > is_convertible<> derives from mpl::bool<>, however it seems > that a bug > > in the compiler prevents it from converting to > mpl::true_/mpl::false_ > > in the above scenario (and in the program attached) > > > > A workaround for this is to change it to the following: > > > > f(mpl::true_); > > f(mpl::false_); > > > > f(mpl::bool_<is_convertible<>::value>()); > > > > Presumably this will be present in other places but this is > the only > > one our developers have hit so far. > > > > Thanks, > > > > Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

3 2

uBLAS problem under Borland C++builder
by 徐亮 05 Oct '04

05 Oct '04

Hi Boost people I'm starting with uBLAS matrix class under Borland C++ builder. I copied a sample from boost document and complied it. There are always mistakes in "traits.hpp". I tried again in VC and no problem occured. I just wonder what's wrong with BC. I have copied the boost dir into BC's Include dir and have added it in Project->Options->Directories. Here's my copied code: #include <boost/numeric/ublas/matrix.hpp> #include <boost/numeric/ublas/io.hpp> int main () { using namespace boost::numeric::ublas; matrix<double> m (3, 3); for (unsigned i = 0; i < m.size1 (); ++ i) for (unsigned j = 0; j < m.size2 (); ++ j) m (i, j) = 3 * i + j; std::cout << m << std::endl; } Thanx 　　　　　　　

1 0