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

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