April 2009 - Boost-users - lists.preview.boost.org

BGL: Debug too slow
by John Krajewski 16 Apr '09

16 Apr '09

It seems my debug build is exteremely slow in the parts that call boost graph library functions. Is there a way I can force the boost lib files to compile with optimizations even though the rest of my app is in debug?? Cheers -John Krajewski

3 2

tss.hpp:82: undefined reference to 'boost::detail::get_tss_data
by wuguangjun 16 Apr '09

16 Apr '09

Hello boosters: When I build boost_1_38, I use 'configure -prefix=/usr' command to put boost files in /usr/ directory. I compile Lintel a paser for cello99, a error occurs: the information is as follows: /usr/include/boost-1_38/boost/thread/tss.hpp:82: undefined reference to 'boost::detail::get_tss_data(void const*)' Is it a bug for boost or some other causes? Would anyone help me? 2009-04-15 wuguangjun

2 3

[coroutine] Bus error on Mac OS X 10.5
by Nat Goodspeed 16 Apr '09

16 Apr '09

I've downloaded the not-yet-Boost coroutine library [1] from the Boost Vault [2]. On a couple of Linux platforms I was able to build and run example/banana.cpp -- though I had to patch noreturn.hpp for gcc 3.3 [3]. On an Intel "fall 2008" MacBook Pro running OS X 10.5.6, with Apple gcc 4.0.1, I had to comment out the MAP_ANONYMOUS mmap() flag [4] -- I don't think that flag is available in Darwin. (Obviously that patch should test for MAP_ANONYMOUS rather than simply suppressing it.) Having done that, the 'banana' example program built -- but I get a bus error trying to run it. I did find an exchange on the Apple mailing list [5] in which Kevin Van Vechten recommends #defining _XOPEN_SOURCE before #include <ucontext.h>: >> Why the bus error? What am I doing wrong? > This is a known issue where getcontext(3) is writing past the end of > the ucontext_t struct when _XOPEN_SOURCE is not defined > (rdar://problem/5578699 ). As a workaround, define _XOPEN_SOURCE > before including ucontext.h. Unfortunately, adding such a #define [6] does not resolve my bus error. The article [7] in which I found ref [5] proposes a classic-C program [8] to illustrate use of getcontext() and swapcontext(). On my Mac, that program produces a seg fault with or without #define _XOPEN_SOURCE. So my apologies: this may very well not be a Boost issue, or even a pre-Boost issue. I'm very excited to try the coroutine library, but crashing on Macs is a show-stopper for us. Has anyone else succeeded in running it on a Mac? [1] http://www.crystalclearsoftware.com/soc/coroutine/index.html [2] http://www.boostpro.com/vault/index.php?action=downloadfile&filename=boost-… [3] attached boost-coroutine.gcc33.patch [4] attached boost-coroutine.osx-mmap.patch [5] http://lists.apple.com/archives/darwin-dev/2008/Jan/msg00232.html [6] attached boost-coroutine.osx-context.patch [7] http://hjiang.net/archives/220 [8] attached swapcontext_test.c *** boost/coroutine/detail/noreturn.hpp~ Sun Aug 20 10:11:09 2006 --- boost/coroutine/detail/noreturn.hpp Fri Mar 27 17:00:00 2009 *************** *** 40,48 **** --- 40,54 ---- #if defined(__GNUC__) + #if __GNUC_MAJOR__ > 3 || (__GNUC_MAJOR__ == 3 && __GNUC_MINOR__ > 3) #define BOOST_COROUTINE_NORETURN(function) \ function __attribute__((__noreturn__)) \ /**/ + #else // gcc 3.3 + #define BOOST_COROUTINE_NORETURN(function) \ + function + /**/ + #endif // gcc 3.3 #elif defined (BOOST_MSVC) *** boost/coroutine/detail/posix_utility.hpp~ Sun Aug 20 13:11:09 2006 --- boost/coroutine/detail/posix_utility.hpp Fri Mar 27 18:52:17 2009 *************** *** 129,135 **** void * stack = ::mmap(NULL, size, PROT_EXEC|PROT_READ|PROT_WRITE, ! MAP_PRIVATE|MAP_ANONYMOUS, -1, 0 ); --- 129,135 ---- void * stack = ::mmap(NULL, size, PROT_EXEC|PROT_READ|PROT_WRITE, ! MAP_PRIVATE /*|MAP_ANONYMOUS*/, -1, 0 ); *** boost/coroutine/detail/context_posix.hpp~ Sun Aug 20 13:11:09 2006 --- boost/coroutine/detail/context_posix.hpp Fri Mar 27 20:57:44 2009 *************** *** 42,47 **** --- 42,56 ---- * NOTE2: makecontext and friends are declared obsolescent in SuSv3, but * it is unlikely that they will be removed any time soon. */ + // NOTE3 (per http://lists.apple.com/archives/darwin-dev/2008/Jan/msg00232.html) + // > Why the bus error? What am I doing wrong? + // This is a known issue where getcontext(3) is writing past the end of the + // ucontext_t struct when _XOPEN_SOURCE is not defined (rdar://problem/5578699 ). + // As a workaround, define _XOPEN_SOURCE before including ucontext.h. + // [While we could make that OSX-specific, it seems harmless enough?] + #ifndef _XOPEN_SOURCE + #define _XOPEN_SOURCE + #endif #include <ucontext.h> #include <boost/noncopyable.hpp> #include <boost/coroutine/exception.hpp> // From http://hjiang.net/archives/220 #include <signal.h> #include <stdio.h> // NOTE (per http://lists.apple.com/archives/darwin-dev/2008/Jan/msg00232.html) // > Why the bus error? What am I doing wrong? // This is a known issue where getcontext(3) is writing past the end of the // ucontext_t struct when _XOPEN_SOURCE is not defined (rdar://problem/5578699 ). // As a workaround, define _XOPEN_SOURCE before including ucontext.h. // [While we could make that OSX-specific, it seems harmless enough?] #ifndef _XOPEN_SOURCE //#define _XOPEN_SOURCE #endif #include <ucontext.h> // Signals which routine should run. volatile int g_turn; void routineOne(ucontext_t* self, ucontext_t* other) { int numbers[] = {1, 2, 3}; int i; for (i = 0; i < sizeof(numbers)/sizeof(int); ++i) { printf("Routine one: %d\n", numbers[i]); // Call other with current continuation if (g_turn != 1) { g_turn = 1; swapcontext(self, other); } } } void routineTwo(ucontext_t* self, ucontext_t* other) { int numbers[] = {-1, -2, -3}; int i; for (i = 0; i < sizeof(numbers)/sizeof(int); ++i) { printf("Routine two: %d\n", numbers[i]); if (g_turn != 2) { g_turn = 2; swapcontext(self, other); } } } int main() { // Continuations ucontext_t cont_one; ucontext_t cont_two; ucontext_t cont_main; // one stack for each thread char stack_one[SIGSTKSZ]; char stack_two[SIGSTKSZ]; // Initialize the coutinuations. cont_one.uc_link = &cont_main; cont_one.uc_stack.ss_sp = stack_one; cont_one.uc_stack.ss_size = sizeof(stack_one); cont_two.uc_link = &cont_main; cont_two.uc_stack.ss_sp = stack_two; cont_two.uc_stack.ss_size = sizeof(stack_two); getcontext(&cont_one); makecontext(&cont_one, (void (*)())routineOne, 2, &cont_one, &cont_two); getcontext(&cont_two); makecontext(&cont_two, (void (*)())routineTwo, 2, &cont_two, &cont_one); g_turn = 0; // Call routineOne with current continuation. Continue from here // after routineOne finishes. getcontext(&cont_main); if (g_turn == 0) { setcontext(&cont_one); } return 0; }

3 6

Strange issue with boost::date_time
by Jared Lee Richardson 15 Apr '09

15 Apr '09

I'm having a strange issue with the boost::date_time library - My program runs fine with both release and debug compilations, but will not run in standalone / the command line. I.e., the program works when I run it through the MS visual studio debugger, whether the program is built with debugging symbols or not. It also ran correctly when run through dependency walker. It does not run when run without either of those, i.e. on a command line. Through print statements I determined that it is running through the initialization and crashes(windows brings up the crashed executable dialog, "This.exe has encountered a problem and needs to close.") The exact code it crashes on is the allocation of a class that contains time_durations: class TempClass { public: list<time_duration> TDList; time_duration SomeOtherDuration; }; and then crashes when allocating this object: TempClass* pTempClass; pTempClass = new TempClass; // <--- Crashes when run without debugger. The program crashes on the new. I tried putting a try/catch block around it for any exception to see if any were being hit and none are. I had tried to reproduce this issue in a much simpler program, but have not had any luck thus far - the test program always runs correctly. Any ideas or things to try? This is not the first instantiation of a time_duration(pushed into a vector a few lines before this), but may be the first instantiation of a std::list. The pointer listed above is actually a member of another class that is contained in a vector, if that matters(shouldn't). Any help? If not I can try again at creating a simpler program that exhibits the problem and paste it. Thanks, Jared

5 9

Windows linking problem
by John Dlugosz 15 Apr '09

15 Apr '09

I got an error cannot open file 'libboost_program_options-vc80-mt-1_38.lib' The file is called 'boost_program_options-vc80-mt-1_38.lib' without the extra prefix. According to the documentation, that's as it should be. So, why is the autolink feature not getting the name right? I'm not the first one to try this, for sure! --John TradeStation Group, Inc. is a publicly-traded holding company (NASDAQ GS: TRAD) of three operating subsidiaries, TradeStation Securities, Inc. (Member NYSE, FINRA, SIPC and NFA), TradeStation Technologies, Inc., a trading software and subscription company, and TradeStation Europe Limited, a United Kingdom, FSA-authorized introducing brokerage firm. None of these companies provides trading or investment advice, recommendations or endorsements of any kind. The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer.

4 6

[TortoiseSVN] Please upgrade if using TortoiseSVN 1.6.0
by Beman Dawes 15 Apr '09

15 Apr '09

There is a bug in TortoiseSVN 1.6.0 that can cause a log request to loop endlessly under certain conditions. When this happens an endless stream of requests is sent to the server. This may be why the svn.boost.org server is getting hammered, and response time is slow. It affects trac too, as it runs on the same server. The problem is fixed in release 1.6.1, so please upgrade to that release if you are currently running TortoiseSVN 1.6.0. Thanks, --Beman

1 0

Help with custom smart pointer const correctness
by Dominique Devienne 15 Apr '09

15 Apr '09

Given the class design below, and the ObjectPtr smart pointer implementation, I can't instantiate ObjectPtr<T> with T = const Car because a const Car returns a shared_ptr<const CarIdent> from the ident() methods, not a shared_ptr<CarIdent> as currently implemented (see compile error at the end of this post. VS2005). >From my very limited meta-programming experience, I'm guessing I need to fix the current typedef shared_ptr<ident_type> ident_ptr; in ObjectPtr to somehow use MPL to "return" shared_ptr<typename T::ident_type const> given a const T, and shared_ptr<typename T::ident_type> given a T, but I don't know how to do that yet. If someone would be kind enough to point me in the right direction, I'd appreciate. Thanks, --DD #include <boost/static_assert.hpp> #include <boost/type_traits/is_base_of.hpp> #include <boost/shared_ptr.hpp> #include <boost/cast.hpp> #include <string> using namespace std; using namespace boost; class Object; class Ident { public: Object* object() const { return object_; } protected: Ident() : object_(0) {} ~Ident() {} friend class Object; void set_object(Object* e); private: Object* object_; }; typedef shared_ptr<Ident> IdentPtr; typedef shared_ptr<const Ident> ConstIdentPtr; class Object { public: ConstIdentPtr ident() const { return ident_; } const IdentPtr& ident() { return ident_; } protected: Object(shared_ptr<Ident> ident) : ident_(ident) { ident_->set_object(this); } virtual ~Object() { ident_->set_object(0); } private: shared_ptr<Ident> ident_; }; void Ident::set_object(Object* obj) { BOOST_ASSERT((!object_ && obj) || (object_ && !obj)); BOOST_ASSERT(!obj || obj->ident().get() == this); object_ = obj; } template <class T> shared_ptr<typename T::ident_type> ident_of(T* t) { BOOST_STATIC_ASSERT((is_base_of<Object, T>::value)); return static_pointer_cast<typename T::ident_type>(t->ident()); } template <class T> shared_ptr<typename T::ident_type const> ident_of(const T* t) { BOOST_STATIC_ASSERT((is_base_of<Object, T>::value)); return static_pointer_cast<typename T::ident_type const>(t->ident()); } template <class T> class ObjectPtr { public: typedef T object_type; // TODO remove constness typedef typename T::ident_type ident_type; typedef shared_ptr<ident_type> ident_ptr; // TODO: make const correct public: // this Ctor fails to compile if passing a const Car for example. // TODO: figure out (MPL? enable_if?) the correct ident_ptr type // based on the constness of T explicit ObjectPtr(T* obj) : ident_(ident_of(obj)) {} T* get() { return polymorphic_downcast<T*>(ident_->object()); } T* operator->() { get(); } T& operator*() { *get(); } ident_ptr ident() const { return ident_; } // TODO: Safe-Bool idiom // TODO: implicit conversion from ObjectPtr<T> to ObjectPtr<const T>? private: ident_ptr ident_; }; struct CarIdent : public Ident { string make, model; CarIdent(const string& make, const string& model) : make(make), model(model) {} }; class Car : public Object { public: typedef CarIdent ident_type; Car(const string& make, const string& model) : Object(shared_ptr<CarIdent>(new CarIdent(make, model))) {} }; void const_correct_smart_ptr(int /*argc*/, const char* /*argv*/[]) { ObjectPtr<Car> z(new Car("Nissan", "350Z")); ident_of(z.get())->model = "370Z"; // TODO: fix ident_of for smart pointers to avoid .get() // This fails to compile with: error C2664: // 'boost::shared_ptr<T>::shared_ptr(const boost::shared_ptr<T> &)' : // cannot convert parameter 1 from 'boost::shared_ptr<const Car::ident_type>' // to 'const boost::shared_ptr<CarIdent> &' // instantiated from explicit ObjectPtr(T* obj) : ident_(ident_of(obj)) {} ObjectPtr<const Car> zz(new Car("Ford", "Mustang")); }

1 1

Re: [Boost-users] [Boost.Test] workaround for reported memory leaks when using MFC?
by gast128 15 Apr '09

15 Apr '09

> Unfortunately, when the MFC dll gets unloaded prior to shutdown it invokes > the CRT's memory leak detection routine which falsely reports that any > memory allocated by the (yet to be unloaded) unit test framework has been > leaked. Obviously, the problem here is with MFC and searching the 'net will > yield numerous complaints on the topic. we have the same problem with our normal app. The leak is reported due to a destructor call of a global object. One way to tweak the reporting, is to extend the lifetime of the global object. This can be done by explicitly linking the main application with mfc (on vc++ 7.1: project properties, linker, input, additional properties mfc71d.lib). Make sure to force a relinking / rebuild after setting this. Not sure if this helps for u.

1 0

[parameter] Composing ArgumentPacks
by er 15 Apr '09

15 Apr '09

Hi All, Two argument packs can be composed with the comma operator. However, if each ap is built using positional arguments, it doesn't seem to work, at least as attempted in example (5) below. Any help would be appreciated. BOOST_PARAMETER_KEYWORD(tag, x) BOOST_PARAMETER_KEYWORD(tag, y) template<class ArgumentPack> void f(ArgumentPack const & args){ double x_val = args[x]; double y_val = args[y]; std::cout << "x = " << x_val << ", y = " << y_val << std::endl; } int main() { typedef boost::parameter::parameters< parameter::required<tag::x>, parameter::required<tag::y> > spec_xy_t; double x_val = 9.0; double y_val = 0.1; f((y=y_val,x=x_val)); // (1) f(spec_xy_t()(x=9.0,y=0.1)); // (2) f(spec_xy_t()(x_val,y_val)); // (3) f( ( (y = y_val), (x = x_val) ) ); // (4) //OK so far spec_x_t spec_x; spec_y_t spec_y; f( ( spec_x(x_val),spec_y(y_val)) // (5) ); //no match operator[<unnamed>::x] return 0; }

1 1

Re: [Boost-users] Strange issue with boost::date_time
by Neil Groves 15 Apr '09

15 Apr '09

On Tue, Apr 14, 2009 at 11:55 PM, Marshall Clow <marshall(a)idio.com> wrote: > At 3:20 PM -0700 4/14/09, Jared Lee Richardson wrote: > >> I'm going to try to remove even more code and narrow it down, and then >> I'll paste the offending code. Basically all I'm doing right now is >> initializing some time variables, then opening and reading from a >> file. It crashes on allocating 5 chars inside a function that reads >> from a file(or on allocating the class that contains a >> std::list<double> and a double). >> > > Um - do you have a bad pointer somewhere, or a buffer overflow? > When I see "Crashes while allocating", I think "heap corruption". > -- > -- Marshall > This is definitely my experience also. In fact if it is the only explanation for an access violation from a non-overloaded new operator that I am aware of. Hence I would be trying to find the perpetrator of the heap corruption. I suggest adding: #include <crtdbg.h> ... _CrtSetDbgFlag(_CRTDBG_CHECK_ALWAYS_DF); http://msdn.microsoft.com/en-us/library/5at7yxcs(VS.71).aspx This typically causes the offending code to be close to the region where the debugger stops. Often this finds the problem within minutes, however I have experienced problems with assertions attaching to the JIT debugger. Hence you may still be in trouble since you state that the debugger must be detached for the problem to occur. My hope is that the fault does occur with a debugger attached, but there are no obvious symptoms and that increasing the CRT checking will bring the problem to the fore. I hope this helps. Best wishes, Neil Groves

2 3