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

RE: [Boost-users] Re: Question
by Sliwa, Przemyslaw (London) 16 May '05

16 May '05

This returns in both cases the same value. Is there a way around this problem at all? Thanks Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Eric Niebler Sent: 12 May 2005 18:13 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: Question Sliwa, Przemyslaw (London) wrote: > Thanks for help, > > I have an additional question. Does anyone know why the following > program gives different output on VC++ 6 and VC++ 2003 NET? It is pretty > strange, isn't it? > <snip> VC6 is pretty strange, I agree. In particular, it doesn't handle functions with non-type template parameters well. The following code should demonstrate the problem. (Hopefully I'm remembering the problem correctly -- I don't have VC6 installed on this machine.) template<int I> void foo() { std::cout << I << '\n'; } int main() { foo<1>(); foo<2>(); return 0; } If you must use VC6, avoid non-type template parameters for function templates. -- Eric Niebler Boost Consulting www.boost-consulting.com _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

3 2

shared_ptr, intrusive_ptr ...
by Jean-François Brouillet 15 May '05

15 May '05

I'm confused again :( I've read the note about shared_ptr & this in the documentation, but I'm still a bit lost as to how I'm supposed to overcome the problem. The code below would work fine if I were to remove all this smart pointer business and handle new & delete by hand. But because this kind of code is just so typical of the huge Java library I'm porting, there's not much choice but to get an as "surprise free" object model as possible. As far as I understand what is going on, the reason why the runtime is reporting a "double free" is because two independently constructed shared_ptrs' do not share the same reference count, which intrusive_ptr would allow me to do, except that intrusive_ptr doesn't vary "polymorphically" with its body as shared_ptr does. Obviously, I could "roll my own" solution by modifying shared_ptr to use a common reference count the way intrusive_ptr does, but I was wondering: is that my only course of action? Or has anyone found a workaround/clever use of the boost::xxx_ptr to overcome this situation? Many thanks -- JFB #include <iostream> #include <boost/shared_ptr.hpp> struct TaskDoneStruct { virtual void success() = 0 ; virtual void failure() = 0 ; } ; typedef boost::shared_ptr<TaskDoneStruct> TaskDone ; struct LongTaskStruct { TaskDone callback ; virtual void setCallback(const TaskDone& td) { this->callback = td ; } virtual void run() { start() ; complete() ; } virtual void start() = 0 ; virtual void complete() { if (callback) { callback->success() ; } } virtual ~LongTaskStruct() {} } ; struct MyTaskStruct : public LongTaskStruct, TaskDoneStruct { MyTaskStruct() { setCallback(TaskDone(this)) ; // this is the root of the // double free problem } virtual void start() { std::cout << "Starting lengthy process ..." << std::endl ; } virtual void success() { std::cout << "Done lengthy process ..." << std::endl ; } virtual void failure() { std::cout << "Aborted lengthy process ..." << std::endl ; } } ; typedef boost::shared_ptr<MyTaskStruct> MyTask ; void tastTest() { MyTask myTask(new MyTaskStruct) ; myTask->run() ; } int main( int argc , char * argv) { std::cout << "Tests starting ..." << std::endl ; tastTest() ; std::cout << "Tests done ..." << std::endl ; return 0 ; } /* BoostTests(1270) malloc: *** Deallocation of a pointer not malloced: 0x300180; This could be a double free(), or free() called with the middle of an allocated block; */

3 11

Re: shared_ptr, intrusive_ptr ...
by Gottlob Frege 15 May '05

15 May '05

> Date: Sun, 15 May 2005 12:00:55 +0100 > From: Jean-Fran?ois Brouillet <verec(a)mac.com> > Subject: [Boost-users] Re: shared_ptr, intrusive_ptr ... > To: boost-users(a)lists.boost.org > Message-ID: <A42134A2-8F76-4A80-956A-D0070056100F(a)mac.com> > Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed > > On 2005-05-15 00:49:54 +0100, "Peter Dimov" <pdimov(a)mmltd.net> said: > > > Anyway, here's my revised snippet that satisfies all my criteria: > -- > JFB > > #include <iostream> > > #include <boost/intrusive_ptr.hpp> > > struct ObjectStruct { > int refCount ; > > ObjectStruct() : refCount(0) { > std::cout << std::hex << (int) this << " - I'm born" << > std::endl ; > } > > void refer() { ++refCount ; } > void unrefer() { if (--refCount == 0) delete this ; } > > virtual ~ObjectStruct() { > std::cout << std::hex << (int) this << " - I'm gone" << > std::endl ; > } > } ; > > void intrusive_ptr_add_ref(ObjectStruct * p) { p->refer() ; } > void intrusive_ptr_release(ObjectStruct * p) { p->unrefer() ; } > Are you going to make it thread-safe (or is that something you just left out of the example?)?

1 0

[ANN][fsm][statechart] New interface available
by Andreas Huber 15 May '05

15 May '05

During the review it was decided to change the interface of the fsm submission as follows: - Library name and namespace are changed to statechart - Reactions are specified with a member typedef in the state type instead of a type that is passed as template argument A version with the above changes applied can be found in the sandbox under the statechart directories and here: http://boost-sandbox.sf.net/Statechart.zip The docs can also be browsed online: http://boost-sandbox.sf.net/libs/statechart I have not yet had the time to document the new reaction specification interface but it should be obvious how to use it from the examples and the tests. Feedback is highly welcome. Thanks & Regards, Andreas P.S. Future posts regarding this library will no longer carry the tag [fsm] but only [statechart] -- Andreas Huber When replying by private email, please remove the words spam and trap from the address shown in the header.

1 0

looped typedefs and/or class definitions
by martin f krafft 14 May '05

14 May '05

Some of my code includes the following (abbreviated): typedef Edge<Node0000Type, Node0000Type> Edge00000000Type; typedef Edge<Node0000Type, Node0001Type> Edge00000001Type; typedef Edge<Node0000Type, Node0010Type> Edge00000010Type; typedef Edge<Node0000Type, Node0011Type> Edge00000011Type; //... you can imagine where this is going. Instead of 32 typedefs, I was wondering if it weren't possible to use the compiler to achieve the above. Similarly, let's say I wanted to define a set of empty classes: class AB : public A, B {}; class AC : public A, C {}; class BC : public B, C {}; can I use the compiler to automate this? Thanks for any comments. -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net "arguments are extremely vulgar, for everyone in good society holds exactly the same opinion." -- oscar wilde

2 1

[mpl] calling a function depending on type or instance as argument
by martin f krafft 14 May '05

14 May '05

Dear all, another MPL question (at least I think it's the domain of MPL): I have a templated class Foo and need to make an instance based on the type of an instance at various points in the code, which is simple: template <typename T> Foo<T> get_foo_for(T const&) { return Foo<T>; } //... get_foo_for(myObj).doSomethingWithFoo(); At times, I don't have an instance of an object, just the typename. I do not want to assume that all types are default constructible (which could be expensive anyway), so I have to do: Foo<T>().doSomethingWithFoo(); I am hiding these two behind two separate macros which I would like to unify. The situation is actually more complex but boils down to the following question: How can I create a macro (#define) such that its definition does different things when passed an instance or a typename. Something like: if is_type(Arg) do_this(); else do_that(); Again, I cannot figure this out from the MPL docs. There is enable_if and if_c, but both seem to deal with the definition of symbols and/or types, not with the selection of functions to call. I appreciate any hint. -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net "arguments are extremely vulgar, for everyone in good society holds exactly the same opinion." -- oscar wilde

1 0

[mpl] calling a function depending on availability of *this
by martin f krafft 14 May '05

14 May '05

Dear all, I am trying to create an assertion macro which can be used inside and outside of class member functions. When inside a class, I want it to be able to use information from *this, which obviously does not exist outside of a class member function. Right now, I have two different macros and I would like to unify the API. The situation is actually more complex but boils down to the following question: How can I create a macro (#define) such that its definition does different things when inside a class member function than outside. Something like: if exists(*this) assertion(a, *this); else assertion(a); such that this is resolved to either of the assertion() overloads at compile time. I guess this is the domain of MPL, but my head just doesn't work in MPL-terms (yet). I would appreciate if someone gave me a solution to this case and promise to try to work my way into MPL more with it. -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net "arguments are extremely vulgar, for everyone in good society holds exactly the same opinion." -- oscar wilde

1 0

enable_if< is_convertible > w/ explicit constructor trouble
by me22 14 May '05

14 May '05

After reading a recent Joel on Software article ( http://www.joelonsoftware.com/items/2003/10/13.html ) and the discussion it generated, I decided to try to write some sort of general strong typedef. I've started out by creating a specific class similar to what would be generated by a `typedef strong_typedef<tag,wrapped_type> name;`, and am trying to use enable_if and is_convertible to allow explicit construction from any type that can be converted to the wrapped type. Unfortunately, I keep getting errors such as the following: main.cpp:36: error: no matching function for call to `position::position(int)' main.cpp:7: error: candidates are: position::position(const position&) main.cpp:14: error: position::position() Compiler version: g++ (GCC) 3.3.5-20050130 (Gentoo Linux 3.3.5.20050130-r1, ssp-3.3.5.20050130-1, pie-8.7.7.1) Here's the cut down code. I'd greatly appreciate it if someone could point me towards what I'm doing wrong. Hopefully I'm just missing something elementary. ( I'm quite new to these boost features. ) On a side note, has anyone come up with a good method of wrapping these rather long SFINAE calls? Thanks, - Scott McMurray // begin code // #include <boost/type_traits.hpp> #include <boost/utility/enable_if.hpp> class position { public: typedef position my_type; typedef unsigned wrapped_type; private: wrapped_type m_data; public: position() {} // This works fine explicit position(boost::enable_if< boost::is_convertible<int, wrapped_type>, int>::type f) : m_data( f ) {} /* // But if I try this, I get the errors template <typename fT> explicit position(typename boost::enable_if< boost::is_convertible<fT, wrapped_type>, fT>::type f) : m_data( f ) {} */ }; int main(int argc, char* argv[]) { position x = position(5); } // end code //

3 4

numeric_cast
by Marcin Kalici�ski 13 May '05

13 May '05

How do I make boost::numeric_cast work with a custom class? What templates do I need to specialize? converter? numeric_cast itself? The documentation is not quite clear about that, I have read it and studied the source code for some time, but found no clue. I think there should be a section about that in the docs. cheers, M.

2 1

Serialization of cross-pointing objects
by Henrik Jacobsson 13 May '05

13 May '05

Hi! I want to serialize objects that link to each other through pointers, like a linked list, but in this case a more complex graph with several different classes and proxies (a graph which can contain loops and and null-pointers). But I get the "pointer conflict" exception since the serialization discovers that an object have been serialized once before through its pointer. The contents should be serialized/deserialized only once, and then the pointers should just be kept consistent. An alternative is of course to replace all the pointers with indeces which are memory-independent. But that's a messy affair. In the copy constructor I solved something similar through a map<Type*,Type*> which mapped from the old pointers to the newly allocated ones, so that they could be properly replaced. Is something like this hidden somewhere in the lib? Any suggestions? Best regards, Henrik Jacobsson

3 15