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

boost::thread_group design flaw
by Pete 17 Mar '04

17 Mar '04

I'm pretty convinced that there is a serious design flaw in boost::thread_group or at the very minimum, the documentation should better address the issue. Let's take a look at the thread_group.cpp example provided with boost::thread: #include <boost/thread/thread.hpp> #include <iostream> int count = 0; boost::mutex mutex; void increment_count() { boost::mutex::scoped_lock lock(mutex); std::cout << "count = " << ++count << std::endl; } int main(int argc, char* argv[]) { boost::thread_group threads; for (int i = 0; i < 10; ++i) threads.create_thread(&increment_count); threads.join_all(); } This example works fine until I make the following small change to main() which breaks it: int main(int argc, char* argv[]) { boost::thread_group threads; for (int j = 0; j < 2; ++i) { for (int i = 0; i < 10; ++i) threads.create_thread(&increment_count); threads.join_all(); } } The above code will break the second time threads.join_all() is called. The problem is that all boost::thread objects created in the first iteration survive the join_all() call. Worse still, the 2nd join_all() call sends another join() message to each thread it created but boost::thread::join() does not check the m_joinable flag and simply tries to free the system resources it previously freed again causing a system fault. To correct the problem above, I can change the code as follows: int main(int argc, char* argv[]) { boost::thread* pt[10]; boost::thread_group threads; for (int j = 0; j < 2; ++i) { for (int i = 0; i < 10; ++i) pt[i] = threads.create_thread(&increment_count); threads.join_all(); for (int i = 0; i < 10; ++i) threads.remove_thread(pt[i]); } } While this code no longer causes a GPF, I do not believe that the boost::thread objects created by boost::thread_group::create_thread() actually ever get destroyed. I have set a break point in boost::thread::~thread() but this breakpoint is never reached. If I physically delete each pt[i] after removing it from the thread_group I'm getting an assertion error telling me that the pointer I'm trying to delete is not a valid debug-heap pointer, but this error will most likely disappear if I stop using the debug heap (I haven't tried yet). As a boost::thread_group user, I made the following assumptions: - a boost::thread created through boost::thread_group::create_thread will be removed and cleaned up properly when a) the thread terminates or b) the thread_group is deleted - after returning from a call to join_all() the boost::thread_group has no boost::thread object members - I need not specifically keep track of boost::thread objects created by boost::thread_group and destruction of these objects is encapsulated within boost::thread_group I now find that none of these assumptions is quite correct. Perhaps I good immediate remedy would be to implement a boost::thread_group::clear() method which would throw an error if any of the threads where still in a running state and else would clear its threads collection and destroy the objects? How about thread objects "remembering" their membership in a thread_group so that the thread group can be informed when a thread terminates?

3 3

boost::thread problem
by Pete 17 Mar '04

17 Mar '04

I'm having a problem with my boost multithreaded application. My app runs fine as a Windows Service, but as a Unix daemon it stops responding after some time. I noticed that the fewer threads I have available, the quicker the application stops responding: HP-UX(64) after 20mins, RH8(256) after >1hr, Linux RH9(1024) >5hrs. This led me to assume that perhaps the threads are not cleaned up properly. Therefore, I wrote a little test app which ought to demonstrate the problem. To my surprise, the test application showed an entirely different problem, but again, only under Unix (it works as expected under Windows). I'm totally lost what the reason is and appreciate any help. Expected: Actual: --------------- --------------- Start running Start running Start swimming Finish running Start cycling Start swimming Finish swimming Finish swimming Finish cycling Start cycling Finish running Finish cycling In windows I get the actual output but under Linux RH9, I get the Actual output. Under RH9, the first five lines of the actual output appear immediately, but the last one takes approx. 4secs. In other words, only the last threads don't sleep the number of secs specified. #include <string> #include <list> #include <iostream> #include <unistd.h> #include <boost/thread/thread.hpp> class MyThread { protected: std::string m_strName; int m_iDuration; public: MyThread(const std::string & s, int i) : m_strName(s), m_iDuration(i) {} void operator()() { std::cout << "Start " << m_strName << std::endl; sleep(m_iDuration); std::cout << "Finish " << m_strName << std::endl; } }; int main(int argc, char *argv[]) { std::list<MyThread*> lst; std::list<MyThread*>::iterator itr; MyThread* pT; boost::thread_group tg; lst.push_back(new MyThread("running", 6)); lst.push_back(new MyThread("swimming", 2)); lst.push_back(new MyThread("cycling", 4)); for (itr = lst.begin(); itr != lst.end(); itr++) { pT = *itr; tg.create_thread(boost::ref(*pT)); } tg.join_all(); while (!lst.empty()) { delete lst.front(); lst.pop_front(); } return 1; }

2 1

Trouble building libbgl-viz.a
by Steve Juranich 16 Mar '04

16 Mar '04

Here's my system specs (let me know if I'm leaving something out): SuSE Linux 8.2 gcc v. 3.3 barbet (juranich)$ g++ -v Reading specs from /usr/lib/gcc-lib/i486-suse-linux/3.3/specs Configured with: ../configure --enable-threads=posix --prefix=/usr --with-local-prefix=/usr/local --infodir=/usr/share/info --mandir=/usr/share/man --libdir=/usr/lib --enable-languages=c,c++,f77,objc,java,ada --disable-checking --enable-libgcj --with-gxx-include-dir=/usr/include/g++ --with-slibdir=/lib --with-system-zlib --enable-shared --enable-__cxa_atexit i486-suse-linux Thread model: posix gcc version 3.3 20030226 (prerelease) (SuSE Linux) flex v. 2.5.4 bison v. 1.75 Description of problem: When I cd into $BOOST/libs/graph/src and type 'make', I get the following errors: barbet:/local/src/boost_1_31_0/libs/graph/src # make flex -Pbgl_undir_ -ographviz_graph_lex.cpp graphviz_lex.ll bison -p bgl_undir_ -d -v -o graphviz_graph_parser.cpp graphviz_parser.yy graphviz_parser.yy: warning: 1 shift/reduce conflict g++ -ftemplate-depth-50 -DGRAPHVIZ_DIRECTED=0 -I../../.. -g -c graphviz_graph_lex.cpp graphviz_lex.ll: In function `int bgl_undir_lex(yystype*)': graphviz_lex.ll:85: error: request for member `ptr' in `*lvalp', which is of non-class type `int' graphviz_lex.ll:87: error: request for member `ptr' in `*lvalp', which is of non-class type `int' make: *** [graphviz_graph_lex.o] Error 1 So, line 85 in graphviz_lex.ll looks like this: I've looked at the source with a friend who (unlike myself) actually has some experience with lex and yacc, but he couldn't spot the problem either. I know that the caveat at the top of the file says that the makefile probably won't work on my system, so are there any suggestions on what I could do to make it work? Thanks a lot. -- Stephen W. Juranich Science Applications Intl. Corp. (SAIC) (520) 570-7706

2 4

RE: [Boost-users] VC 7.1 & boost.thread
by Ben Hutchings 16 Mar '04

16 Mar '04

raja <raja(a)genesyslab.com> wrote: > Hi, > I am getting the following error. I see that it is trying to link to > the static library, but I still get this error. It's not trying to link to the static library - you are. > Would appreciate if someone can correct my usage. > Linking to lib file: lib"boost_thread"-vc71-mt-sgd-1_31.lib > > c:\Boost\include\boost-1_31\boost\config\auto_link.hpp(269) : > fatal error > C1189: #error : "Mixing a dll boost library with a static runtime is a > really bad idea..." <snip> In your project's properties, under C/C++ > Code Generation, set the Runtime Library to be "Multi-threaded DLL" in the release configuration and "Multi-threaded Debug DLL" in the debug configuration. If you are not using the Visual Studio IDE then use the /MD and /MDd compiler switches respectively.

1 0

General problem on compiling a boost regex code...
by Benjamin Kaiser [gamigo AG] 16 Mar '04

16 Mar '04

Hello, I'm using the boost library for the first time, and I get a compiler error. :-( I don't know why, because I think the boost library is installed successful (I have tested the SuSE rpm file and the download version from boost.org) I'm using gcc 3.x and boost 1.30 on SuSE Linux 9.0... Header file: ------------------------------------------- #include <string> #include <fstream> #include <syslog.h> #include <boost/regex.hpp> #include <vector> using namespace boost; ------------------------------------------- CPP file: ------------------------------------------- ... regex expression("([0-9]+)(\\-| |$)(.*)"); ... ------------------------------------------- Thats all I do alluded to the boost regex library... if I want to compile the files, I get the following compiling error: ------------------------------------------- #g++ -ansi -pedantic-errors -Wall -o statd ./projekt/statd.cpp compiling C++ using -ansi -pedantic-errors -Wall -o statd /tmp/cc2KYvxs.o(.text+0x1951): In function `Accounting::SaveTraffic()': : undefined reference to `boost::reg_expression<char, boost::regex_traits<char>, std::allocator<char> >::reg_expression[in-charge](char const*, unsigned, std::allocator<char> const&)' /tmp/cc2KYvxs.o(.text+0x1994): In function `Accounting::SaveTraffic()': : undefined reference to `boost::reg_expression<char, boost::regex_traits<char>, std::allocator<char> >::~reg_expression [in-charge]()' collect2: ld returned 1 exit status ------------------------------------------- It would be nice if someone can help me with this problem... I'm new at C++ and don't know how to fix this. :-( If I only include the regex.hpp and do the "using namespace boost" everything is compiling fine. Only if I do the "regex expression(...)" in the CPP File I get the errors... :-( Many thanks for help! If someone will help me via ICQ: 20320644 Mit freundlichen Grüßen / With best regards Benjamin Kaiser Systemprogrammierer gamigo® AG -Deutschland - Butterstr. 13 48431 Rheine Tel. +49 5971-899060 Fax +49 5971-8990611 bkaiser(a)gamigo.de (DH/DSS) PGP-Key auf Anfrage / PGP-Key on request! -------------------------------------------------- Diese E-Mail enthält vertrauliche und / oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet. 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 unauthorised copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

1 0

Re: [Boost-users] regex with msvc-stlport
by Juan Casanueva 16 Mar '04

16 Mar '04

Hi John, I removed BOOST_REGEX_DECL from re_custom_error_messages, re_ten_w, and re_zero_w and It now works perfectly thanks! Juan >Date: Mon, 15 Mar 2004 13:49:51 -0000 >From: "John Maddock" <john(a)johnmaddock.co.uk> >Subject: Re: [Boost-users] regex with msvc-stlport1 >To: <boost-users(a)lists.boost.org> >Message-ID: <03ff01c40a94$68740cd0$ce810252@fuji> >Content-Type: text/plain; charset="iso-8859-1" > > > I have installed stlport 4.6 and tryed to compile regex but I get the >same > > error. > > > > it creates the .lib fine, but then give some unresolved externals when > > creating the dll.. > > what are your setting in regex\user.hpp??? > >Nothing don't change it. > > > if anyone has any idea, I would apreciate it > > thanks > >The only thing I notice is that the errors are only coming from the release >dll build (and not the debug version), is that correct? > >If so it explains why I can't reproduce it, as I only have the standard >edition of vc6 available to me (which doesn't create optimised builds). > >I also notice that the symbols that the linker is complaining about, >probably don't need to be exported anyway, so can you go into both >c_regex_traits.cpp and w32_regex_traits.cpp and remove the BOOST_REGEX_DECL >specifier from the following identifiers: > >re_custom_error_messages >re_ten_w >re_zero_w > >and see if that fixes the issue? If it does be sure to let me know so that >I can patch up the source, > >Thanks, > >John. > > > >------------------------------ > >_______________________________________________ >Boost-users mailing list >Boost-users(a)lists.boost.org >http://lists.boost.org/mailman/listinfo.cgi/boost-users > >End of Boost-users Digest, Vol 144, Issue 2 >******************************************* _________________________________________________________________ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail

2 1

random library fibonacci bug
by Martin Wartens 16 Mar '04

16 Mar '04

There seems to be a bug in the fibonacci random generators. Only the smaller lagged_fibonacci generators work. Code Sample (tested with VC7.1) follows. The debugger says: Run-Time Check Failure #0 - The value of ESP was not properly saved across a function call. This is usually a result of calling a function declared with one calling convention with a function pointer declared with a different calling convention. //lagged_fibonacci4423 and smaller works, //>=9689 fails typedef boost::lagged_fibonacci9689 rng_type; rng_type rng; rng.seed((unsigned)time( NULL )); boost::uniform_int<size_t> uni(0,len-1); //crashes on this line. boost::variate_generator<rng_type, boost::uniform_int<size_t> > die(rng, uni);

1 0

RE: [Boost-users] boost::thread and std::system
by Ben Hutchings 16 Mar '04

16 Mar '04

Jerry Swan <jeremiah.s(a)ntlworld.com> wrote: > I'd like to be able to issue a call to std::system that will time-out > after a specified period. Is it possible to achieve this using the > boost::threads library? You can stop waiting for it to complete, but you can't terminate the external program. > If so could someone kindly sketch how this might be achieved? You need to have the calling thread create a synchronised flag (bool with condition, protected by mutex) and a child thread that calls system() and then sets the flag when that returns. The original thread then performs a timed wait for the flag to be set. Here's my solution. The synchronised flag can be encapsulated as an "event" like this: struct one_time_event { public: one_time_event() : flag(false) { } void trigger() { boost::mutex::scoped_lock lock(mutex); flag = true; cond.notify_all(); } void wait() { boost::mutex::scoped_lock lock(mutex); while (!flag) cond.wait(lock); } bool timed_wait(boost::xtime & time) { boost::mutex::scoped_lock lock(mutex); while (!flag) if (!cond.timed_wait(lock, time)) return false; return true; } private: boost::mutex mutex; boost::condition cond; bool flag; }; (This is a basic event that you use once and then throw away. More complex event classes are possible but harder to use correctly.) Using this class it's really very easy: void call_system(const char * command, one_time_event & event) { std::system(command); event.trigger(); } bool system_with_time_limit(const char * command, boost::xtime & time) { one_time_event event; boost::thread thread( boost::bind(call_system, command, boost::ref(event))); return event.timed_wait(time); } Ben.

2 1

RE: [Boost-users] Re: Using shared_ptr across shared library boun dary
by Ferdinand Prantl 16 Mar '04

16 Mar '04

> From: David Abrahams [mailto:dave@boost-consulting.com] > > "Jim.Hyslop" <Jim.Hyslop(a)Leitch.com> writes: > > It's been my experience > > (with Visual Studio, at least) that allocating memory in > > one DLL and > > deallocating it in another leads to problems. Sometimes it > > doesn't go > > noticed, but when it does, it's really noticeable. > > The library cleverly avoids that, provided the first > shared_ptr to take ownership of the memory does so in the > same DLL where the object was allocated. Actually, I came into the same problem as Jim - I had to ensure the same version of the msvcrt and moreover its debug/release versions to get the code working well. All classes in checked_delete.hpp, smart_count.hpp and shared_ptr.hpp are either templates or completely in headers. The operations with new/delete and allocators are translated/inlined by the compiler into every unit, IMHO. I could not find the trick in the implementation, how to force the deletion of the pointer by the same runtime, if new() was performed in one DLL and delete() in another one. Could you, please, point me to the right direction to the explatation? Thanks, Ferda

4 3

Newbie Q: Boost Graph stuff
by Ori 16 Mar '04

16 Mar '04

Hi everyone, I have a question concerning the Boost Graph library, specifically with strongly connected components. My problem is this: I have a graph that I need to collapse into strongly connected components, then topologically sort the resulting DAG of components. Then I want to traverse the sorted DAG (in order), and for each component (now just a vertex I guess), look up into the original graph to find all the vertices that belong to this component. The application here is a rigid body simulator, and I'm constructing a graph of contact points between bodies. The idea is to find an ordering to apply collision impulses, which logically means from the ground up. Vertices are thus bodies, and edges denote a contact (collision) case. I've constructed the original graph, and using the example online, performed the strong_components algorithm on it. However, I'm unsure of how to correctly create a graph of the components. I presume using the root_map is somehow involved, but I'm kind of at a loss. Here's what I got: // typedefs, most stolen from graph introduction at boost.org typedef std::pair<int, int> Edge; struct pen_index_t { typedef edge_property_tag kind; }; typedef property<pen_index_t, int> PenProperty; typedef adjacency_list<vecS, vecS, directedS, no_property, PenProperty> Graph; typedef adjacency_list<setS, vecS, directedS, no_property, PenProperty> ComponentGraph; typedef graph_traits<Graph>::vertex_descriptor Vertex; ... // actual code Graph contactGraph(NUM_BODIES); ... various junk to fill up the graph ... // now, collapse strongly connected components vector<int> component(num_vertices(contactGraph)); vector<Vertex> root(num_vertices(contactGraph)); bgl_named_params< Vertex*, vertex_root_t, no_property> rootMap = root_map(&root[0]); int num = strong_components(contactGraph, &component[0], rootMap); // create a new graph with each component as a vertex... // ...how? I tried variations on get(...) on the rootMap and such, but was met with the usual nightmarish stl/boost compiler errors when you mess up something involving templates : ) Any help someone out there can give is really appreciated. Thanks! ~ Ori

2 2