- Boost-users - lists.boost.org

[TR1] Using Boost.TR1 installed in /usr/include with GCC
by Anthony Foglia 28 Feb '10

28 Feb '10

We're running into problems compiling the boost math library with GCC's TR1. The problem comes from Boost.TR1, and I believe the #include_next preprocessor directive. We are using the Ubuntu packages for g++ and for Boost 1.40 which installs the headers in /usr/include/boost. Here's a simple example which fails to compile: ---- #define BOOST_HAS_TR1_TUPLE 1 #include <boost/tr1/tuple.hpp> int main() { return 0; } ---- $ g++ -I/usr/include -c -o tr1_test.o tr1_test.cpp In file included from tr1_test.cpp:2: /usr/include/boost/tr1/tuple.hpp:13:43: error: no include path in which to search for tr1/tuple ---- From the docs, defining BOOST_TR1_DISABLE_INCLUDE_NEXT should work, but does not. I have tried inserting the following lines before the #include, to no effect. #define BOOST_TR1_DISABLE_INCLUDE_NEXT #define BOOST_TR1_GCC_INCLUDE_PATH /usr/include/c++/4.4 The only workaround I could find is to explicitly include boost/tr1/detail/config.hpp and undef BOOST_HAS_INCLUDE_NEXT. Am I missing something in the documentation? What is the proper solution? PS: I see a similar problem mentioned over a year ago: <http://lists.boost.org/Archives/boost/2008/12/146552.php>. But would the changes mentioned have been included in 1.40? I see no obvious changes between trunk and 1.40 for the tuple and config headers, though I haven't diffed them. -- Anthony Foglia Princeton Consultants (609) 987-8787 x233

2 1

Re: [Boost-users] [serialization] class versioning changes in boost 1.42 [patch]
by Brian Wood 28 Feb '10

28 Feb '10

Robert Ramey writes: >David Raulo wrote: > >> - this is not mentionned anywhere in the documentation or in the API. >> - you have at least one user who lost backward archive compatibility >> because of this. >> >> See below a patch against svn. Can we please discuss its advantages >> and drawbacks? >> - what good does it do: obviously in my case, restore backward >> compatibility. > > Why can't you just patch your own copy? > >From a comedic standpoint I think that's a good answer. >> - it gives more flexibility to the versioning scheme. Two usefull >> such schemes were described previously in the discussion, where >> classes still have increasing integer versions, but are not possible >> with 8 bits storage. >> >> - what downsides does appying this patch have? Maybe occuring a >> slight overhead on 16-bits platforms? If true, can this be actually >> measured? >> - Would this patch cause any regression? Break any user code which was >> working fine before? Break user archive backward compatibilty? >> > > Would this not break compatibility with binary_?archive ? > Currently binary archive stores the version as a 16 bit > integer. Maybe it wouldn't but it's another thing that would > have to be considered. Even if it didn't, this would break > the "guarentee" that any serialization which works for one > archive class is guarenteed to work with any other one. I've wondered if Robert deliberately misspells guarantee in order to emphasize that there is no guarantee. -- Brian Wood Ebenezer Enterprises http://www.webEbenezer.net (651) 251-9384

3 3

Re: [Boost-users] TR1 unordered_set: compilation failure with initialization
by Aman Sinha 28 Feb '10

28 Feb '10

I want to pass in my own hash functor and comparator to an unordered_set initialization. It fails compilation on g++ 4.2.1, boost version 1.42. It's not clear why, since a similar initialization of an STL set container seems to work fine... see below for the code and compilation failure. Any insight would be highly appreciated. #include <set> #include <boost/tr1/unordered_set.hpp> class foo_compare { public: foo_compare(bool is_gt) : m_is_gt(is_gt) {} bool operator() (int a, int b) { if (m_is_gt) return a > b; else return a < b; } bool m_is_gt; }; class foo_hash1 { public: foo_hash1() {} ~foo_hash1() {} size_t operator() (int val) const { return val % 19; } }; using namespace std; typedef set<int, foo_compare> MySetT; typedef std::tr1::unordered_set<int, foo_hash1, foo_compare> MyUnorderedSetT; size_t hash_func_1(int val) { return val % 19; } class Foo { public: Foo(bool is_gt) : m_set(foo_compare(is_gt)), // this initialization is OK m_unordered_set(foo_hash1() , // this one fails compilation! why ? foo_compare(is_gt)) {} MySetT m_set; MyUnorderedSetT m_unordered_set; }; int main() { bool is_greater_than = true; Foo myfoo(is_greater_than); return 0; } g++ my_unordered.cpp -I ~/temp/boost_1_42_0 my_unordered.cpp: In constructor Foo::Foo(bool)â: my_unordered.cpp:43: error: no matching function for call to âboost::unordered_set<int, foo_hash1, foo_compare, std::allocator<int> >::unordered_set(foo_hash1, foo_compare)â /home/asinha/temp/boost_1_42_0/boost/unordered/unordered_set.hpp:175: note: candidates are: boost::unordered_set<T, H, P, A>::unordered_set(boost::unordered_detail::move_from<boost::unordered_set<T, H, P, A> >) [with T = int, H = foo_hash1, P = foo_compare, A = std::allocator<int>] /home/asinha/temp/boost_1_42_0/boost/unordered/unordered_set.hpp:121: note: boost::unordered_set<T, H, P, A>::unordered_set(const boost::unordered_set<T, H, P, A>&, const A&) [with T = int, H = foo_hash1, P = foo_compare, A = std::allocator<int>] /home/asinha/temp/boost_1_42_0/boost/unordered/unordered_set.hpp:115: note: boost::unordered_set<T, H, P, A>::unordered_set(const A&) [with T = int, H = foo_hash1, P = foo_compare, A = std::allocator<int>] /home/asinha/temp/boost_1_42_0/boost/unordered/unordered_set.hpp:110: note: boost::unordered_set<T, H, P, A>::unordered_set(size_t, const H&, const P&, const A&) [with T = int, H = foo_hash1, P = foo_compare, A = std::allocator<int>] /home/asinha/temp/boost_1_42_0/boost/unordered/unordered_set.hpp:43: note: boost::unordered_set<int, foo_hash1, foo_compare, std::allocator<int> >::unordered_set(const boost::unordered_set<int, foo_hash1, foo_compare, std::allocator<int> >&) Aman Sinha

2 2

complexity problem using multi_index insert()
by Andrew Kokkalis 28 Feb '10

28 Feb '10

Hello, this is my third e-mail regarding multi_index containers. I'm using them for the process of clustering a huge amount of data (posts from an aple blog from the last 4-5 years) I'm counting Euclidean distance between two clusters, which I store in a multi_index container like the above *typedef multi_index_container< cluster_d, indexed_by< ordered_unique< tag<cluster_pair>, composite_key< cluster_d, BOOST_MULTI_INDEX_MEMBER(cluster_d, int, cluster1), BOOST_MULTI_INDEX_MEMBER(cluster_d, int, cluster2)> >, ordered_non_unique< tag<cluster_distance>, BOOST_MULTI_INDEX_MEMBER(cluster_d, float, distance)> > > cluster_index;* My problem is that I have to count 9500! (! = factorial) distances, and store them in the multi_index container. The process of counting distances is costing me 13minutes ( that is because the dimension of each of the two vectors is nearly 1000 or more for which i have to access STL's maps). Another 9 minutes takes the insertion using "std::pair<iterator,bool> insert(const value_type& x);". I would like to know if there is another way to do faster insertion. I've seen you use mailing list for much complicated problems, and I'm sorry for being a noob.But since i chose to use boost I'd like to stick with that, so I'm kindly asking for your help! Thank you in advance Andrew Kokkalis

4 10

[units] conversion_factor in the two directions
by alfC 28 Feb '10

28 Feb '10

Hi, I managed to make CONVERSION_FACTOR definitions to work between to unit systems (atomic and SI). I definied atomic system by following the si example. I can manage to transform lengths (probably because they are base dimensions in both systems, but when I try to do the same for energies (that is a base_dimension only in the atomic system) I can't compile the line with the declaration "en3". The first question is why the first ...CONVERSION_FACTOR doesn't work in the two directions (as it did for lengths). second, why the second ...CONVERSION_FACTOR doesn't help to compile en3. BOOST_UNITS_DEFINE_CONVERSION_FACTOR( atomic::hartree_energy_base_unit, si::energy, double, 4.359744e-18 ); //compiles but doesn't help for en3 BOOST_UNITS_DEFINE_CONVERSION_FACTOR( si::energy, atomic::hartree_energy_base_unit, double, 1./4.359744e-18 ); int main(){ quantity<si::length> l1(1.*si::meter); //works quantity<si::length> l2(1.*atomic::bohr); //works quantity<atomic::length> l3(1.*atomic::bohr); //works quantity<atomic::length> l4(1.*si::meter); //works quantity<si::energy> en1(1.*si::joules); //works quantity<si::energy> en2(1.*atomic::hartree); //works quantity<atomic::energy> en3(1.*atomic::hartree); //work quantity<atomic::energy> en4(1.*si::joules); //does not work, compilation error return 0; } Thank you, Alfredo

2 6

Compile boost for arm-eabi
by Carlos 28 Feb '10

28 Feb '10

Hi How I could compile boost for another arquitecture? I thougth I have done it right, but when compiling my program with a static link to a boost library I'm getting this message: /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /home/carlos/Arm/arm-2009q3/bin/../lib/gcc/arm-none-linux-gnueabi/4.4.1/../../../../arm-none-linux-gnueabi/bin/ld: /usr/local/boost_1_42_0/lib/libboost_system.a(error_code.o): Relocations in generic ELF (EM: 3) /usr/local/boost_1_42_0/lib/libboost_system.a: could not read symbols: File in wrong format collect2: ld returned 1 exit status I'm getting around this problem and I found that it might be because I don't compile boost in the right way. Any ideias how I do that?

3 5

Re: [Boost-users] Compile boost for arm-eabi (Carlos)
by Nikolai Burtsev 28 Feb '10

28 Feb '10

> How I could compile boost for another arquitecture? I know this is not the right way for cross-compile the boost libs, but it works (i have used it for boost-41) cd to boost libraries and create software links there: ln -s /opt/crosstools/bin/arm-linux-g++ g++ ln -s /opt/crosstools/bin/arm-linux-ar ar ln -s /opt/crosstools/bin/arm-linux-ranlib ranlib ln -s /opt/crosstools/bin/arm-linux-as as ln -s /opt/crosstools/bin/arm-linux-ld ld set your PATH export PATH= ". :$PATH" then build it (as it was for gcc)

1 0

[spirit] need advise for parsing a set of integer alternatives
by Zhi Tan 28 Feb '10

28 Feb '10

say I am parsing a message in this format: 2-byte message length + 2-byte instruction code + message body, where the 2-byte instruction code is a predefined set of constants, e.g. 0x01, 0x01234 etc. I could use this parser: short_>>short_>>*char_. and resort to the attributes of short_in a post-check to check whether the instruction code is legal or not . what if I want to embed the legal instruction code set in the parser itself, and determine whether this is a legal message at one go? .i.e. short_>>(what to put here)>>*char? Thanks.

1 0

BOOST_AUTO_TEST_CASE_EXPECTED_FAILURES and BOOST_FAIL
by Rhys Ulerich 27 Feb '10

27 Feb '10

Is there a way to flag that a Boost.Test auto test case is expected to utterly fail per BOOST_FAIL? BOOST_AUTO_TEST_CASE_EXPECTED_FAILURES does not seem to handle this use case correctly. Running Boost.Test 1.42's example 22 [1] returns EXIT_SUCCESS to the operating system. When I replace BOOST_CHECK with BOOST_FAIL within the sample's my_test1 test case the binary returns EXIT_FAILURE to the OS. I have attached the modified version. I see the same behavior for BOOST_REQUIRE_MESSAGE(false,...). The question arises because I'm using BOOST_FAIL within an error handling callback that I provide to a library. When the library issues the callback, all hope is lost and I need to unwind the test case to continue sanely. Thanks, Rhys [1] http://www.boost.org/doc/libs/1_42_0/libs/test/doc/html/utf/user-guide/test…

2 1

unordered_map: user object as key
by André Santee 27 Feb '10

27 Feb '10

I know it has probably been asked a million times here, but I couldn't find any example that I could use. I want to create an unordered_map whose key is an object that I made (vector2), what's the best way of implementing a hash function to it? Can I have an example? struct vector2 { vector2(); vector2(const vector2 &v2); vector2(float fx, float fy); vector2 operator * (float num) const; vector2 operator * (const vector2& v2) const; vector2 operator / (float num) const; vector2 operator / (const vector2& v2) const; vector2 operator - (float num) const; vector2 operator - (const vector2& v2) const; vector2 operator + (float num) const; vector2 operator + (const vector2& v2) const; bool operator == (const vector2& v2) const; bool operator != (const vector2& v2) const; vector2& operator += (const vector2& v2); vector2& operator -= (const vector2& v2); vector2& operator *= (float f); vector2& operator /= (float f); float x,y; }; boost::unordered_map<vector2, int> won't work.

3 4