October 2005 - Boost-users - lists.preview.boost.org

[general/iostreams/iterators] Why not a forward and/or a random access istream iterator?
by Mats Eklund 22 Oct '05

22 Oct '05

Hello, does anybody know whether there is something fundamentally wrong with designing an istream iterator that conforms with the forward, or even random access, iterator concepts? The std::istream_iterator conforms only with the simplest input iterator concept (stream need not support positioning). If a stream supports positioning, I cannot see why this should not work. If a streamable T has variable size on stream, we could do forward iteration, and if T has a known constant size on the stream, we could do random access iteration, right?! Since I haven't found any previous work in this direction on the web, I suspect there is a catch... (performance?) The only disadvantage I see, is that such iterators would have a rather different semantics from the std::(i)stream_iterator:s (equality etc), and that to create an end-iterator in the random access case, we cannot just rely on the default constructor, but need to pass in the stream, perhaps prepositioned at end. The original problem is to log/write data-records T on a persistant device, for which a std::stream(buf) interface is available. A reason for using the streambuf interface is that the log can support wrap-around functionality by simply using an intermediate streambuf that cares for necessary wrap-around functionality only, and otherwise just forwards transported characters to the device's streambuf. An enhanced std::istream_iterator would provide the log user with a convenient interface for reading records T out of the log. Thanks for any hints! /Mats

3 2

Re: [Boost-users] Report from Mont Tremblant C++ Committee meeting
by Thore Karlsen 22 Oct '05

22 Oct '05

On Sun, 9 Oct 2005 21:32:30 -0400, "Beman Dawes" <bdawes(a)acm.org> wrote: [...] >String-algo : Interested, but concern over interface and choice of >functions, generic vs basic_string.5.3 separate proposal Any details on this? I love this library, and I would love to see it standardized in some form. Overall I've found the interface and function choices to be excellent (and I'd love to see even more), and I love that it's generic and not limited to basic_string. I use it on vectors and other containers for different kinds of network protocol parsing. -- Be seeing you.

9 24

New version of mpl_math in the vault; problems compiling...
by Fredrik Hedman 22 Oct '05

22 Oct '05

Hello, I have been playing around with the examples of this fascinating code. This reports on the problems I have encountered so far. The latest examples are compiled with g++ 3.4.4 and '-Wall -Wno-long-long -pedantic -O2' using using boost-1_33 result in a number of errors: 1) The example in runtime_value.cpp gives boost/mpl/math/number_sign.hpp: No such file or directory This files seems to be missing. 2) The example double.cpp get a *lot* of warnings: mpl/math/double_/aux_/integral_exp.hpp:35: warning: comparison between signed and unsigned integer expressions Looking at the offending line it seems that the constants that are shifted should really be unsigned numbers: BOOST_STATIC_CONSTANT( boost::int16_t , exp1 = (number < (1 << (exp16 + exp8 + exp4 + exp2 + 1))) ? 0 : 1 -> BOOST_STATIC_CONSTANT( boost::int16_t , exp1 = (number < (1u << (exp16 + exp8 + exp4 + exp2 + 1))) ? 0 : 1 The same should probably be applied to all the other BOOST_STATIC_CONSTANT in struct integral_exp. Refactor 1u and 65536u into named constants? 3) The example big_integral.cpp complains about extra semicolons. It is the macro BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_APPLY_RESULT in convert_base10_digits.hpp that is the root cause. By commenting out two ';' in the macro the example compiles and runs just fine: #define BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_APPLY_RESULT \ BOOST_STATIC_CONSTANT( \ boost::uint32_t \ , a_0_0 = \ BOOST_PP_CAT( \ base10000_digit_ \ , BOOST_PP_DEC(BOOST_MPL_MATH_BIG_INTEGRAL_MANTISSA_SIZE) \ ) \ ); \ BOOST_PP_REPEAT( \ BOOST_PP_DEC(BOOST_MPL_MATH_BIG_INTEGRAL_MANTISSA_SIZE) \ , BOOST_MPL_MATH_BIG_INTEGRAL_AUX_A_0_MACRO \ , a_0_ \ ) /* ; removed extra */ \ BOOST_PP_FOR( \ (0) \ , BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_PREDICATE_0 \ , BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_NEXT_STATE_0 \ , BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_MACRO_0 \ ) \ public: \ BOOST_PP_REPEAT( \ BOOST_MPL_MATH_BIG_INTEGRAL_MANTISSA_SIZE \ , BOOST_MPL_MATH_BIG_INTEGRAL_AUX_CONVERT_IMPL_MACRO_1 \ , digit_ \ ) /* ; removed extra */ \ /**/ -- Best Regards, Fredrik Hedman

2 3

[Serialization] Loading from archive
by Sam Contapay 22 Oct '05

22 Oct '05

Sorry I tried to post this through the newsserver gateway. I didn't see the message show up after half and hour so I'm assuming it didn't post correctly as this email address wasn't subscribed. I apologize if it does ----------ORIGINAL EMAIL---------------- I've been trying to get loading from an archive to work for a couple of hours now. Maybe I'm missing something so I can use anyones "eyes" for help. This is a program written using VC7.1 and I do have RTTI on. I am able to save data from a MAP file to an ASCII text file using the serialize functions, but I'm having a problem getting that information back into my object. Here are some relevant code snippets if anyone can point me in the right direction it would be really appreciated. Also is their an archive that I can search? This news group is only leting me download like 100 messages. main class with serialize private: std::map<std::string, int> WordMapA; std::map<std::string, int> WordMapB; friend class boost::serialization::access; template <class Archive> void serialize(Archive & ar, const unsigned int version) { ar & WordMapA; ar & WordMapB; } Now here is my save and load save: // Save the bayes filter especially the map for later use std::ofstream ofs("BayesFilter.stb"); boost::archive::text_oarchive outarchive(ofs); outarchive << this; ofs.close(); load: std::ifstream ifs("BayesFilter.stb"); boost::archive::text_iarchive ia(ifs); CObjTest test; ia >> test; ifs.close(); When I save the file it saves to an ASCII file with no problems. When I then try to load the archive back all the Map data variables are still 0 after a call to the load function. I poured over the demo and its pretty much word for word so I don't know whats wrong any help... thanks. Sam

2 2

Bug in XML serialization!!!
by Piyush Kapadia 21 Oct '05

21 Oct '05

Following code results in "unrecognized XML syntax" exception while writing to XML file. However if I change xml_oarchive to text_oarchive or binary_oarchive, it works fine, so I assume it's a bug in xml_oarchive ??? Does any one know how to fix this for xml_oarchive ??? #include <iostream> #include <fstream> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/xml_iarchive.hpp> #include <boost/archive/xml_oarchive.hpp> #include <boost/archive/binary_iarchive.hpp> #include <boost/archive/binary_oarchive.hpp> #include <boost/serialization/vector.hpp> #include <boost/serialization/map.hpp> #include <vector> #include <map> #include <boost/archive/archive_exception.hpp> #include <boost/archive/basic_xml_archive.hpp> class base { protected : std::vector<int> vec; protected : friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & BOOST_SERIALIZATION_NVP(vec); } public : void add(int i) { vec.push_back(i); } int get(int idx) { return vec[idx]; } }; class derived : public base { friend class boost::serialization::access; std::map<std::string, int> hmap; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & BOOST_SERIALIZATION_NVP(boost::serialization::base_object<base>(*this)); ar & BOOST_SERIALIZATION_NVP(hmap); } public : void add(std::string str, int i) { hmap[str] = i; base::add(i); } int get(std::string str) { return hmap[str]; } }; int main(int argc, char* argv[]) { try { derived A; A.add("First", 12); A.add("Second", 34); const derived AA = A; std::ofstream ofs("Test.xml"); assert(ofs.good()); boost::archive::xml_oarchive oa(ofs); oa << BOOST_SERIALIZATION_NVP(AA); ofs.close(); derived B; std::ifstream ifs("Test.xml"); assert(ifs.good()); boost::archive::xml_iarchive ia(ifs); ia >> BOOST_SERIALIZATION_NVP(B); ifs.close(); } catch (boost::archive::xml_archive_exception ex) { std::cerr << " X " << ex.what() << std::endl; } catch (boost::archive::archive_exception ex) { std::cerr << " A " << ex.what() << std::endl; } catch (std::exception ex) { std::cerr << " E " << ex.what() << std::endl; } catch (...) { std::cerr << "Unknown Exception " << std::endl; } return 0; }

3 3

[random] spot the stupidity competition
by Nigel Rantor 21 Oct '05

21 Oct '05

Hail all, Thanks for giving us the libraries, now only if we were all smart enough to use them properley the world would be at peace. As it is I've been having some problems with my own code that is wrapped around the random library. I'm playing around with a game I'm making and needed some randomness, games that are predicatble are just a tad boring... I need to generate numbers with lots of different distributions so creating one or two won't really cut it for me, there would have to be hundreds. Instead I went and basically ripped the random_number_generator described under decorators into two flavours and then created a bunch of overloaded functions that used them. The generators use a single real engine, currently mt19937;. Basically, everythnig works fine, except that calling seed( n ) on the engine doesn't effect the generators that hold a ref to it, which I found somewhat bizzare. So, after a few tweaks I ripped out the code and dumped it into a minimal test case and can see that the seed( n ) method works just fine when I pull raw numbers out of the engine, but still doesn't seem to effect the generators I've defined. I'm fairly sure this is my problem but I was wondering if anyone would be good enough to take a peek and tell me why exactly this isn't working. I did have a search on Gmane but couldn't find anything that looked related to my problem. I'm not sure what is the done thing on this list so I've attached source, if you want a makefile then I can oblige but it is a fairly small case. The test program reads in three numbers, the seed to use, the number of numbers to generate and the max to use when generating numbers from a distribution. Feeding it different seeds makes the raw number change, but not the others. I have also included the output of two runs to show you what I mean. TIA, N seed : 42 num : 10 max : 1000 seeded rng... 10 raw numbers from rng 1608637542 3421126067 4083286876 787846414 3143890026 3348747335 2571218620 2563451924 670094950 1914837113 10 random integers using rng 795 926 666 119 485 903 904 335 227 185 seed : 727364 num : 10 max : 1000 seeded rng... 10 raw numbers from rng 3240174879 3269238853 3044917573 2953574107 2463698974 2129367301 2064401652 2079989189 3644954707 1971100779 10 random integers using rng 795 926 666 119 485 903 904 335 227 185 #include "random.h" int random_num( int max ) { return int_gen( max ); } int random_num( int min, int max ) { return min + random_num( max ); } long random_num( long max ) { return long_gen( max ); } long random_num( long min, long max ) { return min + random_num( max ); } float random_num( float max ) { return float_gen( max ); } float random_num( float min, float max ) { return min + random_num( max ); } double random_num( double max ) { return double_gen( max ); } double random_num( double min, double max ) { return min + random_num( max ); } #ifndef RANDOM_H #define RANDOM_H #include <boost/random.hpp> using namespace boost; typedef mt19937 rng_type; template<class RNG, class IntType = long> class integer_random_number_generator { public: typedef RNG base_type; typedef IntType argument_type; typedef IntType result_type; integer_random_number_generator( base_type& rng ) : _rng( rng ) {} result_type operator()( argument_type n ) { typedef uniform_int<IntType> dist_type; return variate_generator<base_type&, dist_type>(_rng, dist_type(0, n-1))(); } private: base_type& _rng; }; template<class RNG, class RealType = double> class real_random_number_generator { public: typedef RNG base_type; typedef RealType argument_type; typedef RealType result_type; real_random_number_generator( base_type& rng ) : _rng( rng ) {} result_type operator()( argument_type n ) { typedef uniform_real<RealType> dist_type; return variate_generator<base_type&, dist_type>(_rng, dist_type(0, n-1))(); } private: base_type& _rng; }; static rng_type rng; static integer_random_number_generator<rng_type, int> int_gen( rng ); static integer_random_number_generator<rng_type, long> long_gen( rng ); static real_random_number_generator<rng_type, float> float_gen( rng ); static real_random_number_generator<rng_type, double> double_gen( rng ); int random_num( int max ); int random_num( int min, int max ); long random_num( long max ); long random_num( long min, long max ); float random_num( float max ); float random_num( float min, float max ); double random_num( double max ); double random_num( double min, double max ); #endif /* RANDOM_H */ #include "random.h" #include <iostream> using namespace std; int main( int argc, char ** argv ) { unsigned int seed; int num; int max; int i; cin >> ws >> seed; cin >> ws >> num; cin >> ws >> max; cout << "seed : " << seed << endl; cout << "num : " << num << endl; cout << "max : " << max << endl; rng.seed( seed ); cout << "seeded rng..." << endl; cout << num << " raw numbers from rng" << endl; for( i = 0; i < num; i++ ) { cout << rng() << endl; } cout << num << " random integers using rng" << endl; for( i = 0; i < num; i++ ) { cout << random_num( max ) << endl; } return 0; }

2 2

Failures in Solaris/gcc build
by Fernando Mato Mira 21 Oct '05

21 Oct '05

Hello, I built Boost on Solaris 9 with gcc 3.3.3 but I saw many "...failed" messages while building. Also, in the destination lib directory there's no regex library, and I think there should be one. There are only the libraries for: libboost_filesystem libboost_iostreams libboost_prg_exec_monitor libboost_python libboost_signals Thanks

1 0

Boost Test Memory Dumping...
by Matt Amato 20 Oct '05

20 Oct '05

I have a simple Boost program using the unit test framework. When I run the program in debug, I get a dump of memory leaks. I can't find a way to stop these memory leaks from being reported (other than fixing the leaks of course). What is the proper way of disabling this output? I tried using boost::unit_test::unit_test_log.set_threshold_level but that didn't seem to work. Any ideas? Thanks, Matt Running 4 test cases... *** No errors detected Detected memory leaks! Dumping objects -> {185} normal block at 0x00C3ABF0, 32 bytes long. Data: <Oct 19 2005 19:1> 4F 63 74 20 31 39 20 32 30 30 35 20 31 39 3A 31 {182} normal block at 0x00C3AAF0, 32 bytes long. Data: <Oct 19 2005 19:1> 4F 63 74 20 31 39 20 32 30 30 35 20 31 39 3A 31 {176} normal block at 0x00C3A3E0, 976 bytes long. Data: <<!DOCTYPE VAR [<> 3C 21 44 4F 43 54 59 50 45 20 56 41 52 20 5B 3C Object dump complete. Press any key to continue

2 1

[Boost on Pocket PC] Is Boost supported on Pocket PC platforms?
by Ansuman Kar 20 Oct '05

20 Oct '05

Hi, I'm a newbie user of Boost libraries. We have a requirement for making our sources work on Win32 and Pocket PC platforms. From my early look inside the boost libraries I can see the serious advantages of using them; but would like to find out which Boost libraries are supported on Pocket PC before using them. Quick searches through google and on boost.org site did not provide authoritative results, so I'm posting this question here - please let me know if a different alias is better suited for this question. Thanks. -Ansu __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com

3 2

[Boost.Graph] Double sided edge properties in undirected graph
by Ferdi Smit 20 Oct '05

20 Oct '05

What is the easiest (is there any) way to associate an edge property with both edge ends in an undirected graph in the BGL? I would like to iterate over the incident edges of a vertex, and obtain the edge property for that side of the edge the vertex is on. Ie. for A (5) <--E--> (1) B when iterating the incident edges of A, a request for this edge's (E) property would return 5, and when iterating vertex B's incident edge (E) it would return 1. For a directed graph this is simple: just store a "source" property and a "target" property in each edge. However in the case of an undirected graph the source vertex of an edge will always be the one you're iterating over incident edges from. So this wouldn't work, as it always return the "source" property then, no matter if I'm at A or B. A solution that came to mind is to convert the undirected graph into a directed graph with double edges everywhere. Then I can simply iterate over the out_edges and obtain the right property. However, this seems to be tedious and possibly a waste of space as I'm reinventing undirected graphs manually. Plus I have to rewrite edge removal etc. to keep things consistent. I'm aware that this is how BGL works internally anyhow for undirected graphs, so I don't want to reinvent it. Another "solution" is to store an associated vertex with the properties for the edge, giving two (vertex, value) pairs per edge. First of all this wastes space, and secondly it proves difficult to typedef this property inside the graph type as it is a cyclic dependency. External property maps could do the trick, but the space issue is still unresolved. (I'm primarily speaking of adjacency_lists in this post) A different issue: isn't BGLs genericity an illusion? Compared to for example the illusion of container independence in the STL? So far I've always needed to know what containers are used for the graph in order to make assumptions about iterator validity. For any algorithm using removal or addition a list is basically required, with all the added problems of the vertex indices not being ints... Is BGL still being actively developed? Can I expect positive changes in this area in the future? I really like the concept (no pun) of the BGL, but it's difficult to use at the moment imho. -- Regards, Ferdi Smit (M.Sc.) Email: Ferdi.Smit(a)cwi.nl Room: C0.07 Phone: 4229 INS3 Visualization and 3D Interfaces CWI Amsterdam, The Netherlands

1 0

MSVC and PDB files.
by Jonathon Bell 20 Oct '05

20 Oct '05

How do I use BOOST JAM to build and deploy PDB files for both the debug and release configurations of the various BOOST libraries? I'm using the Visual C++ 7.1 Toolset... _________________________________________ //* Jonathon Bell jbell(a)mathsoft.com //* Mathsoft Engineering & Education, Inc //* 101 Main Street, Cambridge, MA, 02142

3 2

Tokenizer: Low value as seperator
by CN 20 Oct '05

20 Oct '05

Hi! Library version 1.33.0. The input string contains null characters. I want to use null character as the seperator. The following code produces only "X". Why "Y" and "Z" are discarded and how to fix this code? ========= #include<iostream> #include<boost/tokenizer.hpp> int main(int argc, char* argv[]) { using namespace std; using namespace boost; string str="X\0Y\0Z"; typedef tokenizer<boost::char_separator<char> > Tok; char_separator<char> sep("\0"); Tok tokens(str, sep); for(Tok::iterator tok_iter = tokens.begin();tok_iter != tokens.end(); ++tok_iter) cout << *tok_iter; return EXIT_SUCCESS; } ========= Thank you in advance! CN -- http://www.fastmail.fm - Access your email from home and the web

4 6

There is a error in the file licence_info.cpp under tools\bcp
by Huifeng Shen 20 Oct '05

20 Oct '05

The original line 18 is (pay attention the correspond the parentheses) "Copyright|\\(c\\)|? I compile bcp with vc71 toolset and fails. I change this line with "Copyright|\\(c\\)|?" It's OK. I also check the code in the CVS, This error is not corrected.

3 4

Regression tests
by John Grabner 20 Oct '05

20 Oct '05

Hi Guys, I remember seeing the regression test laid out in a table on the web somewhere. You could navigate to whichever result you were interested in. can someone point my in the right direction? John.

2 1

Re: [Boost-users] [MultiIndex] Key extraction from a base member
by JOAQUIN LOPEZ MU?Z 20 Oct '05

20 Oct '05

Hello Bruno, thanks for using Boost.MultiIndex! ----- Mensaje original ----- De: Bruno Martínez <br1(a)internet.com.uy> Fecha: Miércoles, Octubre 19, 2005 10:46 pm Asunto: [Boost-users] [MultiIndex] Key extraction from a base member > Hi. > > I have the following problem with MultiIndex. For > > struct Base { > int id; > }; > > struct Der : Base { > }; > > I can't make a key extractor. > [explanation on why it doesnt work snipped] > Maybe the solution is to put some enable_ifs in member. > Basically, something like that should be done, right. Actually enable_if only works for highly compliant compilers, so I'll have to resort to some uglier mechanism, but it can be done. This is planned to be solved (or added as a new feature, if you wish) in Boost 1.34. The problem was first brought to my attention in http://tinyurl.com/dnfzs. In the meantime I'm afraid you are forced to write your own key extractor (which is trivial, in any case.) Best regards, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

question about subgraphs
by Jeffrey Holle 19 Oct '05

19 Oct '05

I am using boost 1.33.0. I want to use a subgraph to represent a spanning tree of a graph. This application requires adding and deleteing edges to just the tree and not the underlining graph though. Note that this wouldn't violate the requirements of a subgraph, a spanning tree contains all the vertices, and a subset of the edges of the underlining graph. Is it possible to do this, maybe through inheritance?

2 1

[MultiIndex] Key extraction from a base member
by Bruno Martínez 19 Oct '05

19 Oct '05

Hi. I have the following problem with MultiIndex. For struct Base { int id; }; struct Der : Base { }; I can't make a key extractor. This way typedef multi_index_container< Der, indexed_by< ordered_unique<member<Der, int, &Der::id> > >Set; doesn't work because &Der::id is of type pointer to member of Base. This other typedef multi_index_container< Der, indexed_by< ordered_unique<member<Base, int, &Base::id> > >Set; doesn't work because of the ChainedPtr overloads in member. I tried casting &Der::id to the appropiate type, but couldn't make it work. Maybe the solution is to put some enable_ifs in member. Regards, Bruno

1 0

boost::any & Serialization
by Piyush Kapadia 19 Oct '05

19 Oct '05

Why can't we use boost::any for serialization ? Isn't there any work around to make this work ???

2 1

[Serialization]class(type) name in xml archive?
by Sean Huang 19 Oct '05

19 Oct '05

Is it possible to add class(type) name in xml archive? It would be a big help when looking at the xml output (yes, we want to display those files). for example (from demo.xml): <s type="bus_schedule" class_id="0" tracking_level="0" version="2"> ^^^^^^^^^^^^^^^^^ Thanks, Sean

2 3

Boost Birds of Feather @ OOPSLA 05?
by Jeff Garland 18 Oct '05

18 Oct '05

All - I'm going to be in San Diego @ OOPSLA next week and I'm wondering if any Boost users and developers would be interested in getting together for a BOF one evening. Obvious topic of discussion is all things Boost/C++/TR2. Right now I'm thinking Tuesday night would be a good time. Let me know if there is interest and I'll set this up. Jeff Note, I cross-posted to the user and developer list b/c the idea is this would be open to everyone with an interest in Boost.

5 7

FW: custom iterator_adaptor::pointer
by Sigg Christian 18 Oct '05

18 Oct '05

Dear boosters, I'm trying to figure out why there is no way to define the iterator_adaptor's pointer type. I'm trying to write an iterator that concurrently accesses two containers, but the interface should look like it's only one container. Therefore, it is necessary to write my own value, reference and pointer type. However, the iterator_adaptors pointer type is always value*. Is it save to just overload the pointer typedef and reimplement operator->()? I'm not too much of a c++ expert, so if anyone could comment on that, I would greatly appreciate it. Christian

2 1

Any way to use boost lib standalone?
by Hendrik Bock 18 Oct '05

18 Oct '05

Hello there! I have searched the boost-users mailing-list for quite some time, but found no answer to my question: Is there a possibility to separate a library from the rest of boost? I only want to use the regexp library of boost and i have to include all related header-files plus libs into our company-cvs. Till now i wasn't able to find out, which *.hpp files are needed by boost and which ones aren't. Kind regards, Hendrik Bock

2 2

Boost Graph Library: Python Bindings
by Yu Huang 17 Oct '05

17 Oct '05

Hi guys, I report a small bug related to bgl.file_kind.adjlist. I tried it using an inputfile in the specified format, but got nothing. And i checked the source code, basic_graph.cpp and found on line 129: case gfk_adjlist: break; which means simply not dealing with it at all. I guess those two guys are so busy to fill some codes in it. Thanks, Yu

5 6

Building with Dinkumware Unabridged Library?
by Drumheller, Michael 17 Oct '05

17 Oct '05

There does not seem to be a build option for using Boost with the Dinkumware Unabridged Library. Yet I know there are Boost users who use Dinkumware. Can someone explain? Thanks,

1 0

[unit test framework] Command-line arguments; documentation
by Paul Giaccone 17 Oct '05

17 Oct '05

How do you access command-line arguments in test cases? According to http://boost.org/libs/test/doc/components/utf/index.html: "This function [boost::unit_test::test_suite* init_unit_test_suite ( int argc, char* argv[] )] should create and initialize top level instance of the class test_suite. The NULL pointer returned by the function is treated as a non-initialized test_suite - testing aborted and boost::exit_test_failure is returned. In other case the framework runs created instance of the class test_suite. The framework forwards command line argument specified during test invocation. It's guarantied that any framework-specific command line arguments are excluded." I understood this to mean that the command-line arguments are passed to the test cases ("The framework forwards command line argument specified during test invocation."), but this doesn't seem to be the case. So how do you pass and access command-line arguments in the test cases? I tried passing them explicitly: void digitise_graph_test(int argc, char* argv[]) { [...] } boost::unit_test::test_suite* init_unit_test_suite(int argc, char* argv[]) { test -> add(BOOST_TEST_CASE(&my_test(argc, argv))); } but this gives a compilation error: error C2102: '&' requires l-value A secondary question, or rather, a polite request: Could I ask that someone edit the documentation for unit test framework, please? While is it mainly intelligible, some of it is difficult to make sense of. Apologies to Gennadiy Rozental: this isn't meant as a personal criticism nor to sound patronising, but when a developer's first language is not the language in which the documentation is to be published, it is preferable to have a speaker of that language write the documentation or at least edit it.. I can cope with the grammatical or spelling errors and am not criticising those, but the meaning is somewhat difficult to derive in some places. Documentation must be understandable to be useful, and to my mind, it is not at the moment. Thanks. Paul Giaccone

3 5