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

boost::shared_ptr templated constructor?
by Lehrer, Joshua 06 May '03

06 May '03

WRT boost::shared_ptr<>, why not template the explicit constructor that takes a T*? This way, it could select the correct destructor at construction time: struct base { ~base(); }; struct sub : public base { ~sub() }; boost::shared_ptr<base>(new sub); The above code currently only calls ~base(); Templating the explicit single argument constructor would allow the above code to call ~sub(). The constructor now takes only a T*, and therefore populates the pointer to the destructor function with a call to ~T(), which is ~base() in the above example. If, however, the constructor took an X*, the above example would pass a sub*, and it could populate the default destructor function with a pointer to ~X(). Comments? Joshua Lehrer FactSet Research Systems NYSE:FDS

3 2

BGL leaky?
by Martin Okrslar 06 May '03

06 May '03

Dear Boost users, is the BGL leaky? For example the code added below produces memory leaks (as seen from also added valgrind output). When I comment out the line marked by /* THIS LINE DOES IT */, there are no leaks. I am running BGL 1.29 and gcc 3.2. Any ideas for a work-around? cheers Martin #include <iostream> // for std::cout #include <utility> // for std::pair #include <algorithm> // for std::for_each #include <boost/graph/graph_traits.hpp> #include <boost/graph/adjacency_list.hpp> #include <boost/graph/dijkstra_shortest_paths.hpp> using namespace boost; int main(int,char*[]) { // create a typedef for the Graph type typedef adjacency_list<vecS, vecS, bidirectionalS> Graph; // Make convenient labels for the vertices enum { A, B, C, D, E, N }; const int num_vertices = N; // writing out the edges in the graph typedef std::pair<int, int> Edge; Edge edge_array[] = { Edge(A,B), Edge(A,D), Edge(C,A), Edge(D,C), Edge(C,E), Edge(B,D), Edge(D,E), }; int num_edges = sizeof(edge_array)/sizeof(edge_array[0]); // declare a graph object Graph* g = new Graph(num_vertices); // add the edges to the graph object for (int i = 0; i < num_edges; ++i) add_edge(edge_array[i].first, edge_array[i].second, *g); /* THIS LINE DOES IT */ delete g; return 0; } okrslar@liebrich:~/molgen/domclust/src/test> valgrind --leak-check=yes --show-reachable=yes ./domclust ==23473== valgrind-1.0.4, a memory error detector for x86 GNU/Linux. ==23473== Copyright (C) 2000-2002, and GNU GPL'd, by Julian Seward. ==23473== Estimated CPU clock rate is 2412 MHz ==23473== For more details, rerun with: -v ==23473== ==23473== ==23473== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0) ==23473== malloc/free: in use at exit: 384 bytes in 1 blocks. ==23473== malloc/free: 3 allocs, 2 frees, 544 bytes allocated. ==23473== For counts of detected errors, rerun with: -v ==23473== searching for pointers to 1 not-freed blocks. ==23473== checked 6470036 bytes. ==23473== ==23473== definitely lost: 0 bytes in 0 blocks. ==23473== possibly lost: 0 bytes in 0 blocks. ==23473== still reachable: 384 bytes in 1 blocks. ==23473== ==23473== 384 bytes in 1 blocks are still reachable in loss record 1 of 1 ==23473== at 0x4004A3A8: __builtin_new (vg_clientfuncs.c:125) ==23473== by 0x4004A3F6: operator new(unsigned) (vg_clientfuncs.c:138) ==23473== by 0x4035E4F1: ??? (bits/stl_alloc.h:482) ==23473== by 0x4035E40D: ??? (bits/stl_alloc.h:532) ==23473== ==23473== LEAK SUMMARY: ==23473== definitely lost: 0 bytes in 0 blocks. ==23473== possibly lost: 0 bytes in 0 blocks. ==23473== still reachable: 384 bytes in 1 blocks. ==23473== -- ----------------------------------------------------------------------------- Martin Okrslar MPI for Molecular Genetics phone: ++ 49 + 30 / 8413-1166 Computational Molecular Biology Fax: ++ 49 + 30 / 8413-1152 Ihnestrasse 73 email: okrslar(a)molgen.mpg.de D-14195 Berlin URL: http://cmb.molgen.mpg.de -----------------------------------------------------------------------------

2 1

More Mac Compiler Woes - Boost 1.30.0 - Thread and type_traits
by Bobby Thomale 05 May '03

05 May '03

Hey guys, I am having some more problems getting Boost 1.30 to compile on the Mac. I am using CW Pro 8.3, compiling a TARGET_CARBON project. I am having trouble getting the MP implementation of threads to compile. It seems to be having trouble with function pointers where the syntax function<type> is used. I remember at one point it was mentioned that: function<type, type...> was deprecated and we were to use either the more universally supported: functionN<type,type...> or the cleaner: function<void myfuncSig()> instead if our compiler supported it. (Mine doesn't support that last one yet I don't think.) I am guessing that the oldest syntax was removed and no one has bothered to update the Mac carbon thread implementaton to compile again since then? That one seems pretty easy to fix - just update to a supported syntax? (I could do this, update to the functionN<type,type...> syntax since that is what my compiler supports. If someone wants me to make and submit the changes to them for inclusion I'd be happy to, if I knew who to send them to...) Or is there a better/easier workaround? Secondly, the Signals library is not compiling. It's actually not a Signals problem, I think it is a type_traits problem. I am getting a bunch of strange compile errors, in a boost::type_traits file called: is_base_and_derived.hpp It looks like there is a template near the top that is the first thing that fails. There is a define that goes into an alternate implementation for certain broken compilers. Perhaps CodeWarrior needs to be added to the list of compilers that compiles the "broken compiler" version of the stuff in this header? Thanks in advance for your help! -- Bobby --------------------------------------------------------------------- Bobby Thomale Senior Software Developer Inoveon Corporation http://www.inoveon.com/ ---------------------------------------------------------------------

5 6

bug in pthreads implementation of recursive_mutex
by Feldman, Yulik 05 May '03

05 May '03

Hi, It looks that there is a bug in .../boost_1_30_0/libs/thread/src/recursive_mutex.cpp. The bug shows in the pthreads implementation of the recursive_mutex when the mutex is used with the "condition" synchronization primitive. Here is the description of the bug: When the client code calls wait() function on a condition sending as a parameter a lock object associated with a recursive_mutex ("m_Condition.wait(Lock)"), the following functions get called: boost::condition::wait(L& Lock) at condition.hpp:91 boost::condition::do_wait(M& mutex) at condition.hpp:137 boost::detail::thread::lock_ops<boost::recursive_mutex>::unlock(Mutex& m, lock_state& state) at lock.hpp:54 boost::recursive_mutex::do_unlock(cv_state& state) The implementation of the last function on "winthreads" and "mptasks" resets the 'm_count' data member to zero. The "pthreads" implementation at recursive_mutex.cpp:399 doesn't reset the 'm_count' to zero. As a result, when the calling thread enters pthread_cond_wait() and the mutex is unlocked by the pthreads the 'm_count' still remains non-zero. Starting from this point, the things may go bad. Consider, for example, that another thread tries to acquire the mutex. The thread calls: boost::recursive_mutex::scoped_lock Lock(m_Mutex) boost::detail::thread::scoped_lock<boost::recursive_mutex>::scoped_lock<boos t::recursive_mutex>(Mutex& mx, initially_locked = true) at lock.hpp:66 boost::detail::thread::scoped_lock<boost::recursive_mutex>::lock() at lock.hpp:76 boost::detail::thread::lock_ops<boost::recursive_mutex>::lock(Mutex& m) at lock.hpp:34 boost::recursive_mutex::do_lock() at recursive_mutex.cpp:307 The last function locks the mutex again, and, if BOOST_HAS_PTHREAD_MUTEXATTR_SETTYPE is defined, increments the counter to 2. Now, since the count is more than 1, the function unlocks the mutex. As a result, the mutex remains unlocked, after the lock function is called. The same problem exists in recursive_try_mutex and recursive_timed_mutex. I've tried to fix the code as follows, which seemed to help in my particular case: ---------------------------------------------------------------------------- ------------------------------------ *** recursive_mutex.cpp.orig Sun Feb 23 19:38:11 2003 --- recursive_mutex.cpp Sat May 3 17:32:44 2003 *************** *** 412,417 **** --- 412,418 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } recursive_try_mutex::recursive_try_mutex() *************** *** 608,613 **** --- 609,615 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } recursive_timed_mutex::recursive_timed_mutex() *************** *** 787,792 **** --- 789,795 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } #elif defined(BOOST_HAS_MPTASKS) ---------------------------------------------------------------------------- ------------------------------------ Any comments? Is this indeed a bug? Is the fix correct? Thanks, Yulik. [Non-text portions of this message have been removed]

3 2

uBlas - matrix comparison A == B -> how-to guide ...
by geert_ceuppens 05 May '03

05 May '03

Sharing my effort in working with ublas ... Forgive me if this is described already somewhere - I could not find it immediately in the doc so I looked around in the code ; ) Matrix comparison is implemented using the equals() method defined in matrix_assign.hpp: bool equals (const matrix_expression<E1> &e1, const matrix_expression<E2> &e2) Code snippet: ublas::matrix<double> A(5,5), B(5,5); // Initialise A somehow init_matrix(A); B = A; // This does not compile ... bool equal = ( A == B ); // But this does (and it works ...) bool equal = ublas::equals ( A, B );

1 0

RE: [Boost-Users] How To Build Boost 1.30 With Dev-CPP?
by bjorn.karlsson＠readsoft.com 04 May '03

04 May '03

> From: firingme [mailto:yg-boost-users@m.gmane.org] > E:\LY\boost_1_30_0>bjam sTOOS=mingw This should read sTOOLS, not sTOOS. Bjorn

5 5

RE: [Boost-Users] Code Fail
by Jim.Hyslop 02 May '03

02 May '03

Hossein Haeri [mailto:powerprogman@yahoo.com] wrote: > There is a tiny cpp file attached to this message > that raises the following exception while running: No, there isn't ;-) The email list strips out attachments. If it truly is tiny, then paste it inline. If it's not-so-tiny, then you can upload it to the files area: http://groups.yahoo.com/group/Boost-Users/files/ -- Jim

1 0

RE: [Boost-Users] Re: Is there a "known issues" page anywhere?
by Jim.Hyslop 02 May '03

02 May '03

Alisdair Meredith [mailto:yg-boost-users@gmane.org] wrote: > "Jim.Hyslop" wrote: > > > There is a "Known Issues" section on the wiki (see the link in the > > auto-appended .sig) which is intended to document these > common "gotchas". I > > haven't looked at it for a while, so I don't know how > up-to-date it is. > > I tried looking there first, and after your message went back for a > second look. > Perhaps I am being dense, but I really can't find it. > Closest match is > the 'build issues' page, which I think is for a different thing. > 'Search' the wiki did not help either. That's the page I meant - CommonBuildProblems: http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?CommonBuildPr oblems Sorry about the confusion. Your problem with the lexical_cast is the precise reason I created the page. > Could you post a link to the precise page you were thinking > about? And > I think a more prominent link on the front page may be useful too. There's one under the "Quick Links" section, but as you say it's not very prominent. Remember, the point of a wiki is that *anyone* can change the information. If you think it deserves a more prominent position, then you are free to give it one :=) -- Jim

1 0

PP library...
by Eric Robert 02 May '03

02 May '03

Hi, I think the preprocessor library could help me here but I'm having problems understanding it. Maybe someone will be so kind to explain it to me a bit. Here is what I want to produce: void foo(char); void foo(int); void foo(unsigned int); .... You got the idea. Something that repeats a list of things i.e. here, types. How can I do that? Thanks! Eric [Non-text portions of this message have been removed]

3 3

Code Fail
by Hossein Haeri 02 May '03

02 May '03

Dear all, There is a tiny cpp file attached to this message that raises the following exception while running: "Project ConstStatProject.exe raised exception class EAccessViolation with message 'Access violation at adress ...'. Process stopped. Use Step or Run to continue." and stops at the following line of the share_ptr library: if(use_count_ == 0 && weak_count_ != 0) boost::throw_exception(boost::bad_weak_ptr()); Can anybody help me? BTW, I'm using BCB6. Thanks, --Hossein __________________________________ Do you Yahoo!? The New Yahoo! Search - Faster. Easier. Bingo. http://search.yahoo.com [Non-text portions of this message have been removed]

1 0

[BGL] Online Graph Algorithms (fwd)
by David Pearce 02 May '03

02 May '03

Hello, I am interested in how boosters think online graph algorithms should be written. I have looked at the connected_components algorithm and feel that it breaks some of the ideas behind generic programming. For example, consider the following: > typedef adjacency_list <vecS, vecS, undirectedS> Graph; > Graph G(6); > add_edge(0, 1, G); > add_edge(1, 4, G); > > // create the disjoint-sets structure, which requires > // rank and parent vertex properties > ... > initialize_incremental_components(G, ds); > incremental_components(G, ds); > > generic_algorithm_which_adds_edges(G); At this point, the connected_components data structures may be out-of-sync. Now, I suppose that we could write a wrapper class which looked like a "Graph" and provides its own add_edge to update "ds" correctly. However, this seems overly cumbersome. It strikes me that a better way of doing this would be to use an abstract base class, like so: > template<class T, ...> > class connected_components : public T { > ... > }; > > template<class T> > pair<T::edge_descriptor, bool> > add_edge(connected_components<T>::vertex_descriptor t, ...) { > // do stuff > return add_edge(static_cast<T::vertex_descriptor>(t), ...); > } Thus, we supply "Graph" as the base class and can then safely pass it to "generic_algorithm_which...". So, does anyone know of any reasons not to do this? Cheers, David J. Pearce

1 0

const shared_ptr<>
by Hossein Haeri 02 May '03

02 May '03

Dear all, I'm a newbie in boost! I want to know what the following means: int a = 5; const boost::shared_ptr<int> sp(&a); Does it mean that the following line is an error? *sp = 7; If so, what are the guarantees? Is there any documented guarantee? (I couldn't find anything in the documentation. Excuseme if I'm careless!) Thanks, --Hossein __________________________________ Do you Yahoo!? The New Yahoo! Search - Faster. Easier. Bingo. http://search.yahoo.com

3 11

RE: [Boost-Users] Is there a "known issues" page anywhere?
by Jim.Hyslop 02 May '03

02 May '03

Alisdair Meredith [mailto:yg-boost-users@gmane.org] wrote: > Is there a place where known problems for libraries/compilers are > stored? > > For instance, trying to use boost/time we hit a problem with MSVC7 and > lexical_cast unless we enable the "wchar_t as a native type" option. > While this may be good practice <g> it is not the default, > and caught us > out (and the error is none-too helpful) > > I can imagine this sort of issue is discovered and re-discovered on a > regular basis, and wonder where to look for such known issues before > using list bandwidth. There is a "Known Issues" section on the wiki (see the link in the auto-appended .sig) which is intended to document these common "gotchas". I haven't looked at it for a while, so I don't know how up-to-date it is. If that issue's not documented there, would you mind adding it? -- Jim

3 2

graph (BGL): bad documentation
by schlumpfi106 01 May '03

01 May '03

I just tried to use the boost graph library to create a spanning tree generation algorithm. After wasting a whole day I gave up and wrote all the graph code myself. I'm quiet upset about the useless online documentation for the graph library. Maybe I have to buy the book to be able to use it? Normally I would rely on some good, instructive examples and start my own program by copy-and-paste. Here, there are only a few examples and they are either too simple or too specific. Most documentation topics provide no examples at all. Sometimes I head to guess through the template-loaded headerfiles because there was no documentation. I failed in creating a graph from a given adjacency matrix. I failed in understanding the "interior property" concept. I failed in printing a graph and its properties to graphviz. To the developers: Please, before you implement anything new -- document the old things.

1 0

RE: [Boost-Users] Regexp Homework :-)
by Ben Hutchings 01 May '03

01 May '03

Simon Brown <simon.brown(a)kns.ch> wrote: > A few days ago I received great help with a regexp question - > now here's > another which I doubt can be done. > > Question: if the text string contains three words AAA BBB AAA > then return > the string (if any) which is repeated two or three times. > > Examples: > AAA AAA AAA return AAA > AAA BBB AAA return AAA > AAA AAA BBB return AAA > AAA BBB CCC return <nothing> > > Can this be done with a regular expression? Match against "(\\w+) (?:\\1 \\w+|\\w+ \\1)|\\w+ (\\w+) \\2" and return the match for whichever sub-expression (1 or 2) was matched. (I tested this with Python not Boost regex, but they should be compatible.)

2 1

Is there a "known issues" page anywhere?
by Alisdair Meredith 01 May '03

01 May '03

Is there a place where known problems for libraries/compilers are stored? For instance, trying to use boost/time we hit a problem with MSVC7 and lexical_cast unless we enable the "wchar_t as a native type" option. While this may be good practice <g> it is not the default, and caught us out (and the error is none-too helpful) I can imagine this sort of issue is discovered and re-discovered on a regular basis, and wonder where to look for such known issues before using list bandwidth. -- AlisdairM

1 0

regexp match question
by Simon Brown 01 May '03

01 May '03

Hi All, Here's a question I can't answer. I'm using regexp to dig out strings from data received via Ham Radio, so the txt often is corrupted. What I want to match is two consecutive callsigns separated by whitespace, for example 'hb9drv hb9drv' but not 'hb9drv hb9drX'. I can define the match condition for a callsign, but can I use the match option {2,3} somehow to match the same callsign two or three times? Effectively match the same word two or three times? Help received with much thanks, Simon Brown, Casa Bergenia, CH-7031 Laax www.hb9drv.ch [Non-text portions of this message have been removed]

3 5

Licence terms
by Ben Hutchings 30 Apr '03

30 Apr '03

Where should I post to ask questions and make suggestions about licence terms for Boost libraries? I am specifically concerned about the large number of slightly different wordings of some of the licences, which I think would all have to be included in documentation of a program using the covered libraries.

1 0

RE: [Boost-Users] const shared_ptr<>
by Tom Matelich 30 Apr '03

30 Apr '03

> From: Hossein Haeri [mailto:powerprogman@yahoo.com] > Sent: Wednesday, April 30, 2003 8:29 AM > > I want to know what the following means: > > int a = 5; > const boost::shared_ptr<int> sp(&a); > I assume you know this, but this code is in error. When the last shared_ptr referencing &a goes away, it will try to delete a, which is illegal. There is a workaround defined at http://www.boost.org/libs/smart_ptr/sp_techniques.html#static , though. ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. -----------------------------------------------------------------------

1 0

ptime and time_duration
by Andre du Toit 30 Apr '03

30 Apr '03

Hi. Since time_duration uses <long> to represent each of the fields (H:M:S:FS), I should be able to add any hour offset to a ptime, as long as the resultant ptime is inside the date/time range representable by boost? With the date ranges of boost being 1400-Jan-1 00:00:00 -> 10000-Jan-1 00:00:00, the number of hours between these dates is (very roughly!) 7.534E7. This is less than the <long> limit on my system, which is +/- 2.147E9. That is good. So; doing the following: ptime t_max( date(10000,Jan,1) ); ptime t_min( date(1400,Jan,1) ); time_duration td = t_max - t_min; std::cout << to_simple_string( td ) << std::endl; should produce something like : 7.534E7:0:0 or if you want to write it out: 7533600:0:0 not? Instead; the output I get is : 224112:19:12. This corresponds to a timespan of roughly 25 years which means I just ran into some overflow/underflow problem somewhere. What gives? What time range is supported by time_duration if it does not use the <long> limits? I have difficulty deducing this information from the header files, the source for the lib is not too helpfull either! What am I missing here? Greetings, Andre.

2 1

running regression tests
by Vaclav Vesely 30 Apr '03

30 Apr '03

It's possible to execute all regression tests so it's possible to execute one test suite: bjam test or bjam filesystem But it's impossible to execute several (but not all) test suits at once: bjam config filesystem Only the last (filesystem) suit is executed. It's possible to improve testing.jam to allow it? Can someone give a bit of advice? Thanks and regards, w

4 5

Re: Re: running regression tests
by Dave Abrahams 30 Apr '03

30 Apr '03

Peter Dimov wrote: > David Abrahams wrote: >> BTW, Vaclav, the place to post about Boost.Build issues is >> jamboost(a)yahoogroups.com. >> >> http://groups.yahoo.com/group/jamboost/ > > ... and the place to announce mailing lists is > > http://www.boost.org/more/mailing_lists.htm Updated.

1 0

boost::spirit multiline comment parsing - how?
by Markus Werle 30 Apr '03

30 Apr '03

Hi! While playing around with the examples of the spirit library I feel like I missed some essential points. The far goal I have is to parse a file containing C/C++ comments and assignments of the form name = value; where I drop the comments and store the assignments in a map<string, string>. If You have done something similar, please share Your knowledge with the rest of us. Below is a modifed version of comments.cpp. Here I only try to output any comment found while dropping the other stuff. Why does the parser drop stuff after the 1st assignment? the output of the code is Parsed C-comment successfully! Matched (27) characters: "//test2 /* test1 */// test Any help appreciated. Also I would like not to gather all commenst in one string ... -------- #include "boost/spirit.hpp" #include <iterator> #include <string> #include <iostream> #include <cassert> #include <fstream> #include <map> #include <cassert> /////////////////////////////////////////////////////////////////////////////// // actor called after successfully matching a single character class actor_string { public: actor_string(std::string &rstr) : matched(rstr) { } void operator() (const char *pbegin, const char *pend) const { matched += std::string(pbegin, pend-pbegin); } private: std::string &matched; }; /////////////////////////////////////////////////////////////////////////////// // actor called after successfully matching a C++-comment void actor_cpp (const char *pfirst, const char *plast) { std::cerr << "Parsing C++-comment" << std::endl; std::cerr << "Matched (" << plast-pfirst << ") characters: "; char cbbuffer[128]; strncpy(cbbuffer, pfirst, plast-pfirst); cbbuffer[plast-pfirst] = '\0'; std::cerr << "\"" << cbbuffer << "\"" << std::endl; } /////////////////////////////////////////////////////////////////////////////// // main entry point int main () { using namespace boost::spirit; ///////////////////////////////////////////////////////////////////////////// char const* pCComment = "//test2\n/* test1 */\n\n// test\nalpha = 3.0;\n\n// beta is soo important\nbeta=5.6;\n\n/* \n\nSometimes comments are\ngood since\nbeta = 4.5;\nis a good choice, too\n\n*/"; typedef skipper_iteration_policy<> iter_policy_t; typedef scanner_policies<iter_policy_t> scanner_policies_t; typedef scanner<char const*, scanner_policies_t> phrase_scanner_t; rule<phrase_scanner_t> cpp_comment, ignored, all; cpp_comment = comment_p("/*", "*/") // rule for C-comments | comment_p("//") // rule for C++ comments ; std::string comment_c; all = ( *cpp_comment[actor_string(comment_c)] | *anychar_p ) ; ignored = (space_p | chlit<char>('\n')); parse_info<> result; result = parse (pCComment, all, space_p); if (result.hit) { std::cerr << "Parsed C-comment successfully!" << std::endl; std::cerr << "Matched (" << comment_c.size() << ") characters: "; std::cerr << "\"" << comment_c << "\"" << std::endl; } else { std::cerr << "Failed to parse C/C++-comment!" << std::endl; } std::cerr << std::endl; } Markus -- Compile time analytic differentiation? Yes, at http://daixtrose.sourceforge.net/

2 1

bjam, targets from another projects
by Vaclav Vesely 30 Apr '03

30 Apr '03

Hi, It's possible to specify dependency on a target from another project tree? I want to write to my project's "Jamfile" something like this: --- Jamfile --- exe test : test.cpp <dll>$(BOOST_ROOT)/libs/threads/build/boost_thread : <include>$(BOOST_ROOT) ; --- Thanks and regards, w

2 1

bjam - several projects and ALL_LOCATE_TARGET
by Vaclav Vesely 30 Apr '03

30 Apr '03

Hi, I would like to ask for help with using bjam. I want to have more project directory trees, for example: C:\src\boost C:\src\my_projects\first C:\src\my_projects\first and I want to compile binaries to separate directories: C:\build\boost C:\build\my_projects\first C:\build\my_projects\second So I have to set different values of the ALL_LOCATE_TARGET environment variable for each project. It's unconfortable to set that value each time I want to inovoke the bjam, so I want to set it in the Jamfile of each project. I mean somethin like this: --- C:\src\my_first_project\Jamfile --- project-root ; ALL_LOCATE_TARGET = C:\build\my_projects\first --- Of course, this doesn't work. Can someone propose the solustion? BTW it would be handy to use the "project-root" directive for this reason: --- C:\src\my_first_project\Jamfile --- project-root my_projects/first ; --- The specified path is ment as relative to the ALL_LOCATE_TARGET. Thanks and regards, w

2 1