- Boost-users - lists.boost.org

more ublas test problems
by Benedikt Weber 18 May '03

18 May '03

When compling the bench* tests for ublas using the jamfiles, both cwpro8 and borland do not seem to know about the function rand(). I think that <stdlib.h> should be included somewhere. MSVC seems to do this implicitely. Furthermore, only debug versions are generated for all tests (test* and bench*). Benedikt

3 7

strange behaviour at setup on windows2000
by Anthony Lichnewsky 17 May '03

17 May '03

I had the following trouble setting up boost 1.29.0 on windows 2000 (using vc7, but that's not the point) the first thing bjam did (the version downloaded from sourceforge) was to erase all the files and folders in tools\ so that no build could be performed... setting those files to read-only was enough but that's an unexpected behaviour. Anthony Lichnewsky

2 3

Signals lib - causes mysterious multiple destructor calls
by Chuck Messenger 17 May '03

17 May '03

Testing out the Boost signals lib, I wrote: ----------- #include <boost/signal.hpp> #include <iostream> using namespace std; using namespace boost; struct X { void operator()(int val) { cout << "X <- " << val << endl; } X() { cout << "X constructor\n"; } ~X() { cout << "X destructor\n"; } }; int main() { { cout << " phase 1\n"; signal<void (int)> sig; cout << " phase 2\n"; X x; cout << " phase 3\n"; sig.connect(x); cout << " phase 4\n"; } cout << " phase 5\n"; return 0; } ------------------- The following output is generated (gcc3.2.2 on Mandrake 9.1): ------------------------ phase 1 phase 2 X constructor phase 3 X destructor X destructor X destructor X destructor X destructor X destructor X destructor X destructor phase 4 X destructor X destructor phase 5 ------------------------ Why the multiple calls to X's destructor? I'd have expected just one. It looks like the signals lib is misbehaving. Can anyone offer an explanation? - Chuck

3 2

lambda library: how to use io manipulators?
by Chuck Messenger 16 May '03

16 May '03

The following compiles: (cout << _1 << "\n")(x); but this doesn't: (cout << _1 << endl)(x); (gcc 3.2.2) I can't figure out how to use an io manipulator within a lambda function. Firstly, how do I do it? (e.g. suppose I want to use flush, for which there is no workaround like using "\n"). Secondly, a suggestion: it would be nice if support was built in for the io manipulators. - Chuck

1 0

is_base_and_derived broken in 1.30.0?
by Jeff Muller 16 May '03

16 May '03

Originally posted this on 5/6, but it never showed up. Here goes again... Hi, The following code: const bool isWS = ::boost::is_base_and_derived<std::wstring,std::wstring>::value; sets isWS false with VS.NET 2003 (_MSC_VER=1310) in version 1.30.0 of the libraries. It is correct with version 1.27.0. Thanks, Jeff

2 1

Using the Python interpreter
by dima_mouromtsev 15 May '03

15 May '03

Hello All! I've the problem with the next code: class MyWindow { ... MyWindow() {} ~MyWindow() {} MyOperation() { Py_Initialize(); handle<> main_module(borrowed(PyImport_AddModule("__main__"))); handle<> namespace(borrowed(PyModule_GetDict(main_module.get()))); try { handle<>(PyRun_String("hello = file('c:\\hello.txt', 'w')\n" "hello.write('Hello world!')\n" "hello.close()", Py_file_input, namespace.get(), namespace.get())); } catch(error_already_set) { ... } Py_Finalize(); } }; When I start this code, that's all right. If I move Py_Finalize() to destructor, the program work, but when I close the program an exeption rise (Py_FatalError("UNREF invalid object")). Next, I exclude Boost classes: ... MyOperation() { PyObject* dict = PyModule_GetDict(PyImport_AddModule("__main__")); PyRun_String("hello = file('c:\\hello.txt', 'w')\n" "hello.write('Hello world!')\n" "hello.close()", Py_file_input, dict, dict); } So the program work well and there is no exeptions at exit. What is my problem? How can I use Boost in right way? Please Help Thanks Dima

2 2

shared_from_this from a destructor
by ostware 15 May '03

15 May '03

(This message is also posted on sourceforge. Please forgive the double posting, but I am still unfamiliar with the community...) version: boost 1.30.0 library: smart pointers I am using the boost::enable_shared_from_this base class from the smart pointer library to implement shared_from_this in my classes. I was amazed to find this class, after hours of thashing trying even to figure out the very problem it solves. So, first off, thanks to the boosters. My problem is that shared_from_this() throws boost::bad_weak_ptr if called from my class's destructor. Is this a design decision? a bug? am I misreading a class requirement? The problem (if it is one...) is that sp_counted_base::release() decrements use_count_ before calling dispose(). Then when my class calls shared_from_this from its destructor (via dispose()), sp_counted_base::add_ref fails (use_count == 0) and throws boost::bad_weak_ptr. I would suggest either setting use_count_ after dispose(), or clarifying in the docs that shared_from_this can't be called from a destructor. (... which would send me back to the drawing board to solve my problem) It would also be handy to have an exception less version of share_from_this which justs returns a null shared_ptr. Thanks for any help. ... Michael Ost Something like this (warning: uncompiled, untested!) ... class T : public enable_shared_from_this { public: T() {} ~T() { shared_ptr<T> ptr = shared_from_this(); // trouble: exception, can't get a shared_ptr } }; main() { shared_ptr<T> ptr = shared_ptr(new T()); ptr.reset(); }

2 3

? Using std::max_element() on a multi_array
by Neil Mackie 15 May '03

15 May '03

I'm looking for advice on using std::max_element() with boost::multi_array. When passing a multi_array to max_element() I was expecting to get an iterator referring to the largest element in return. However my compiler tells me the iterator refers to a sub_array<class RAClass::RAPoint,1> instead, and I don't know why or what to do about it. Its clear that inexperience with multi_array and max_element() is the cause. Is anyone able to shed some light on the subject? Details: ―--- .h file ―--- namespace RAClass { typedef boost::multi_array<RAPoint,2> CELL_TABLE; CELL_TABLE* m_cells; ―--- .cpp file ―--- CELL_TABLE::iterator e = std::max_element(m_cells->begin(),m_cells->end(), RAPoint::lower); // RAPoint::lower() compares two RAPoints int height = e->getHeight(); ―--- compiler output ―--- getHeight' : is not a member of 'sub_array<class RAClass::RAPoint,1>' TIA Neil (J Mackie). By the end of this year the Radiocommunications Agency and all responsibility for regulating the radio spectrum and communications services will become part of Ofcom - the new converged regulator for the communications sector. See the Ofcom transitional website at www.Ofcom.org.uk for more details.

1 0

BGL - misc problems encountered
by Chuck Messenger 15 May '03

15 May '03

I've been testing out the BGL, and have found some small problems. Is this a good forum to post these? If not, please set me straight. I have in mind to post more of these -- in the interest of helping iron out the library. Maybe the developer list would be more appropriate? Or a BGL-specific list, if one exists...? Problem 1: In the documentation, under "Using adjacency_list", subsection "Custom Edge Properties", it shows a method for declaring a custom property tag: > struct flow_t { > typedef edge_property_tag kind; > }; > > struct capacity_t { > typedef edge_property_tag kind; > }; It goes on to say that an alternative method is this: > enum edge_myflow_t { edge_myflow }; > enum edge_mycapacity_t { edge_mycapacity }; > > namespace boost { > BOOST_INSTALL_PROPERTY(edge, myflow); > BOOST_INSTALL_PROPERTY(edge, mycapacity); > } In the rest of the code included in the documentation, it uses the first method. Finally, it says: > The file edge_property.cpp shows the complete source code for this example. However, the code in edge_property.cpp is completely different -- using the second method rather than the first. So, at a minimum, the final sentence is wrong. I suggest the alternative methods should both be shown in the sample code. I made the necessary changes to edge_property.cpp -- here's the diff (against Boost release 1.30.0): 78a79,88 > #define ALTERNATE // comment this out to see an alternate method > > #ifdef ALTERNATE > struct flow_t { > typedef edge_property_tag kind; > }; > struct capacity_t { > typedef edge_property_tag kind; > }; > #else 86c96 < --- > #endif 95c105,109 < property_map<Graph, edge_mycapacity_t>::const_type --- > #ifdef ALTERNATE > typename property_map<Graph, capacity_t>::const_type capacity = get(capacity_t(), G); > typename property_map<Graph, flow_t>::const_type flow = get(flow_t(), G); > #else > typename property_map<Graph, edge_mycapacity_t>::const_type 97c111 < property_map<Graph, edge_myflow_t>::const_type --- > typename property_map<Graph, edge_myflow_t>::const_type 98a113 > #endif 125a141,145 > #ifdef ALTERNATE > typedef property<capacity_t, int> Cap; > typedef property<flow_t, int, Cap> Flow; > typedef adjacency_list<vecS, vecS, bidirectionalS, no_property, Flow> Graph; > #else 129a150 > #endif 163a185,188 > #ifdef ALTERNATE > property_map<Graph, flow_t>::type > flow = get(flow_t(), G); > #else 165a191 > #endif Problem 2: edge_proprety.cpp, as included in Boost release 1.30.0, generates warnings under gcc 3.2.2, since it is missing a couple of typenames. Here's the needed patch: 95c95 < property_map<Graph, edge_mycapacity_t>::const_type --- > typename property_map<Graph, edge_mycapacity_t>::const_type 97c97 < property_map<Graph, edge_myflow_t>::const_type --- > typename property_map<Graph, edge_myflow_t>::const_type - Chuck

1 0

RE: [Boost-Users] Re: find_if and calling a member function
by Tom Matelich 15 May '03

15 May '03

> From: Russell Hind [mailto:yg-boost-users@gmane.org] > > Tom Matelich wrote: > > > > 1) write your own algorithm find_if_equal, then you only > need one bind (~10 > > lines of code separate from your call) > > This is what I was looking for, and I was really wondering (although > didn't say it) if there were such a thing in boost already. > I haven't > found one, but there are quite a few bits of boost I'm not > familiar with > yet. > I don't think boost has any extra general algorithms laying around, unless they're hiding in the details directories. ----------------------------------------------------------------------- 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