- Boost-users - lists.boost.org

RE: [Boost-users] boost::thread and std::system
by Ben Hutchings 17 Mar '04

17 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

BUG: boost::lexical_cast ignores virtual functions.
by John F Meinel Jr 16 Mar '04

16 Mar '04

I found that when I call boost::lexical_cast<std::string>() with an object that has virtual functions, it fails to dereference the virtual functions. It also cannot use an Abstract base type. Attached is example code. I think the problem is simply that the file boost/lexical_cast.hpp, on line 185 has: Target lexical_cast(Source arg) This should be Target lexical_cast(Source& arg) At least, this fixes all of my problems. John =:-> #include <iostream> #include <string> #include <boost/lexical_cast.hpp> struct A { virtual int f() const = 0; }; struct B : A { virtual int f() const { return 1; } }; struct C : B { virtual int f() const { return 2; } }; std::ostream& operator<<(std::ostream& os, const A& a) { return os << a.f(); } int main() { C c; std::string s; std::cout << c << std::endl; //This should print 2 s = boost::lexical_cast<std::string>(c); std::cout << s << std::endl; //This should print 2 s = boost::lexical_cast<std::string, B>(c); std::cout << s << std::endl; //This returns 1, not 2 //This fails because it tries to instantiate an object of type A. Why??? s = boost::lexical_cast<std::string, A>(c); std::cout << s << std::endl; }

1 0

Problems building Boost
by Ulf Johansson 16 Mar '04

16 Mar '04

I've downloaded Boost 1.31.0 and tried to build it using the standard bjam procedure from the getting started document. I have WinXP/Home and the standard edition of VC7.1 (.NET 2003). I've tried both the vc7.1 and the vc7.1-stlport toolsets. In both cases the following libraries are generated: boost_date_time boost_filesystem boost_signals boost_thread libbost_date_time libbost_prg_exec_monitor libbost_regex libbost_signals libbost_test_exec_monitor libbost_unit_test_framework During the build process I get compiler warnings because the standard version of VC won't do advanced optimizations, but otherwise there seems to be no problem. Bjam stops after having generated the above libraries. Shouldn't there be more? What's the difference between those starting with boost and those starting with libboost? Do I have to somehow generate the remaining libraries separately? Thanks!

3 3

VC 7.1 & boost.thread
by raja 16 Mar '04

16 Mar '04

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. 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..." Here is my sample: #define _MSC_VER 1310 #define BOOST_LIB_NAME "boost_thread" #define BOOST_LIB_DIAGNOSTIC "on"// Show library file details. #include <boost/config/auto_link.hpp> #include <boost/thread/thread.hpp> #include <iostream> void hello() { std::cout << "Hello world, I'm a thread!" << std::endl; } int main(int argc, char* argv[]) { boost::thread thrd(&hello); thrd.join(); return 0; }

1 0

[BGL] Determining if a graph has a particular internal property?
by Douglas Gregor 16 Mar '04

16 Mar '04

Is this possible? A well-defined return type from get(p, g) when p is not an internal property of graph g would suffice. Doug

3 2

Re: [Boost-users] regex with msvc-stlport1
by Juan Casanueva 16 Mar '04

16 Mar '04

Hi John, I have installed stlport 4.6 and tryed to compile regex but I get the same error. vc-Link ..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.dll ..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.lib Creating library ..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.lib and object ..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.exp boost_regex-vc6-mt-p-1_31.exp : error LNK2001: unresolved external symbol "?re_custom_error_messages@?%C:\Program" (?re_custom_error_messages@?%C:\Program) boost_regex-vc6-mt-p-1_31.exp : error LNK2001: unresolved external symbol "?re_ten_w@?%C:\Program" (?re_ten_w@?%C:\Program) boost_regex-vc6-mt-p-1_31.exp : error LNK2001: unresolved external symbol "?re_zero_w@?%C:\Program" (?re_zero_w@?%C:\Program) ..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.dll : fatal error LNK1120: 3 unresolved externals "link" /nologo /INCREMENTAL:NO /DLL /subsystem:console /out:"..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.dll" /IMPLIB:"..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.lib" /LIBPATH:"C:\Program Files\Microsoft Visual Studio\STLport-4.6\lib" /LIBPATH:"C:\PROGRA~1\MICROS~2\VC98\lib" @"..\..\..\bin\boost\libs\regex\build\boost_regex.dll\msvc-stlport\release\stlport-cstd-namespace-global\stlport-iostream-on\boost_regex-vc6-mt-p-1_31.CMD" it creates the .lib fine, but then give some unresolved externals when creating the dll.. what are your setting in regex\user.hpp??? if anyone has any idea, I would apreciate it thanks Juan >>Date: Sat, 13 Mar 2004 12:31:07 -0000 >From: "John Maddock" <john(a)johnmaddock.co.uk> >Subject: Re: [Boost-users] regex with msvc-stlport >To: <boost-users(a)lists.boost.org> >Message-ID: <022501c408f7$128de550$1a520252@fuji> >Content-Type: text/plain; charset="iso-8859-1" > > > I am trying to build de regex boost library with msvc and stlport4.6.1, >with > > no success!!! > >I regularly test regex with msvc+STLPort and it works for me with >Boost.1.31.0 (OK, I use STLPort 4.6, but otherwise the same settings as you >are using). > > > I use the following to compile (release version): > >The static libs have built OK, but the dll build failed for some strange >reason, I really don't understand that... > >John. > _________________________________________________________________ STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail

2 1

Problem using Boost with Dinkumware STL library
by Darren Cox 16 Mar '04

16 Mar '04

2 1

Re: [Boost-users] mpl rational_c and fixed_c
by Aleksey Gurtovoy 16 Mar '04

16 Mar '04

Matt Calabrese writes: > A couple of days ago I started work on a library where I need a way to > represent rational numbers and fixed point precision numbers at > compile time, so I started creating a compile-time rational number type > following the style of integral_c in boost::mpl. After completing most > operations, I notices that in the mpl/math directory there were already > the beginnings of rational_c and fixed_c types, however, they are far > from complete (in rational_c there are just two static constants for > numerator and denominator and a self typedef, with no specialized > operations). Are these types still in development or are they > abandoned? They are waiting for somebody to pick them up. To some extent, that already happened with rational numbers. There are a couple of more or less complete implementations laying around: * http://tinyurl.com/2qrpq, 'rational_c' implementation by Hugo Duncan; * http://tinyurl.com/39mjl, "static_rational.hpp" by Matthias Schabel (it's a part of YANL, his dimensional analysis library). The main reason one of them is not in the library yet is that their notation for arithmetic operations leaves something to be desired -- you have to write something like rational_plus<r1,r2> instead of mpl::plus<r1,r2> Besides this issue, AFAIK these are already usable. Of course, if you would like to improve them/contribute more optimal/clean/better in any other regard implementation, please do. As for 'fixed_c', I am not aware of any publicly available implementation. Again, contributions are more than welcome! > I'd like to continue my project using > mpl::rational_c and mpl::fixed_c as I need to make them work with the > mpl integral constants anyways. > > So is development on these templates halted, meaning I should just go > ahead and finish my types, or are the types coming soon enough so I > just should just hold off and wait for the cavalry to arrive? No, please go ahead! Cavalry have enough on their plates already (among other things, a new version of the library is upcoming, with plenty of new exciting and useful stuff like associative sequences), so progress on these two is totally in your hands. -- Aleksey Gurtovoy MetaCommunications Engineering

3 5