October 2007 - Boost-users - lists.preview.boost.org

Pattern-matching by metaprogramming - possible?
by Ovanes Markarian 18 Oct '07

18 Oct '07

I don't think that meta-programming will help check you the values. This can only be the case if the values are available (as const with external linkage) at compile time. If you need the checking pattern to be generated during the compilation than take a look at boost::xpressive. With Kind Regards, Ovanes On 10/18/07, Zeljko Vrba <zvrba(a)ifi.uio.no> wrote: > > On Thu, Oct 18, 2007 at 09:04:39AM +0200, Ovanes Markarian wrote: > > > > Hello Zeljko, > > > > may be another approach? Why not using a template method and static > concepts > > inside a function, which ensure that a parameter is of correct type? > > > Hmm, I didn't quite get you. But I want to match both on *type* and > *value* > as shown in this rule: > > > > > > | (Insn "addq", Reg "%rsp") -> (* do something *) > > > > The type here is (Insn, Reg) [particular types of larger variants]; > each component is also matched on the particular value ("addq" and > "%rsp" strings). > > > > > > | (_, Imm x) -> (* any 1st item; the 2nd must be Imm type, and > > > > the x variable is bound to the value carried by the > 2nd > > > > The type here is doesn't matter and (Imm x) - this is a match only > on type where the variable x gets bound to the value carried by > the variant type. > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

1 0

Pattern-matching by metaprogramming - possible?
by Zeljko Vrba 18 Oct '07

18 Oct '07

A short introduction: the ML language has a variant data type, tuples, and pattern matching construct. For example type operand = Imm of int | Reg of string type instruction = Insn of string | Pseudo of string Now, we can create tuples, e.g. (Insn "push", Reg "%rbp") is of type operand * instruction (a tuple of two typed elements). In function definitions, it is possible to do pattern matching as (it could be written slightly better in ML, but I want to show the match syntactic construct): let f x = match x with | (Insn "addq", Reg "%rsp") -> (* do something *) | (_, Imm x) -> (* any 1st item; the 2nd must be Imm type, and the x variable is bound to the value carried by the 2nd item *) | (_, Reg "%rsp") -> (* any value for 1st item _except_ Insn "addq" which has already been handled in the first alternative - alternatives are matched in order and first matching is taken *) | _ -> (* default case for anything else *) This is only a simplified example; but even this example would be much more ugly if rewritten with if/else statements, even if tuple tiers were used. Boost provides both the safe variant type and the tuple type. Would it be possible to construct pattern matching with syntax similar to the above in C++ ? [Yes, there are also other problems - incomplete matches: ML just throws an exception in that case; or redundant cases - nothing happens.]

2 2

Quasi-random
by Joel C. Salomon 17 Oct '07

17 Oct '07

I've found some mention of this in the archives, but it seems it's never been done. How much work would it be to add the quasi-random sequence generators (faure, halton, hammersley, sobol, &c.) from the code at <http://people.scs.fsu.edu/~burkardt/cpp_src/cpp_src.html>? --Joel

1 0

Problem lboost_filesystem-mgw-mt-s-l_33_l
by question boost 17 Oct '07

17 Oct '07

Hi , compiling a project under Code Blocks I get the following message: ld.exe:cannot find -lboost_filesystem-mgw-mt-s-l_33_l I go there http://www.boost.org/doc/html/jam/building.html I do:C:\Documents and Settings\src\boost\tools\jam\src>build mingw --debug (like explained) A new folder named bin.ntx86.debug is added after I go there http://www.boost.org/doc/html/bbv2/installation.html : 1:(done) 2:I do:C:\Documents and Settings\src\boost\tools\build\v2> bjam --version 3:I open the user-config.jam file (like explained) and I modify the file, the modification is the following: import toolset : using ; # GCC configuration # Configure gcc (default version) using gcc : : "C:\mingw\bin\g++" ; 4:I copy bjam.exe from C:\Documents and Settings\src\boost\tools\build\v2 to C:\Documents and Settings\src\boost\tools\build\v2\example\hello and I run bjam (like explained): C:\Documents and Settings\src\boost\tools\build\v2\example\hello>bjam I get the following message: "C:mingwbing++ n'est pas reconnu en tant que commande interne ou externe, un programme executable ou un fichier de commandes .." ..... "...failled gcc.compile.c++ bin\gcc\debug\hello.o... ...skipped <pbin\gcc\debug>hello.exe for lack of<pbin\gcc\debug>hello.o... ...failed updating one target.. ...skipped one target..." Where is my error (4)? Thanks, David _________________________________________________________________ Votez pour vos séries TV préférées et tentez de gagner un voyage à Hawaï ! http://messengerawards.divertissements.fr.msn.com/

3 2

[ref] tying references between objects
by Robbie Morrison 17 Oct '07

17 Oct '07

Hi all The following kind of code is probably quite common. It involves: * using individual Register objects containing construction parameters to help factory-create new Entity objects * and, in the process, "tying" key object state variables so that each associated Registry object can watch these And yes, I know this breaks encapsulation! I chose C++ references and my code seems quite complicated for what should be a relatively straightforward task. Pointers were rejected because the dereferencing syntax can lead to really obscure numerical expressions (particularly for non-C/C++ programmers). However, boost::ref() seemed to offer a simpler solution and I experimented with this for quite some time to no avail. I also considered using a boost::ptr_vector and 1-tuples. In addition, adding cout statements to ctors and dtors led to more activity than I would have expected. Moreover, there were more dtor calls than ctor calls. Any comments or suggestions gratefully received! Also an explanation of the differences between a "type conversion" operator and an "address-of" operator. Is this simply related to role or is there a distinction? More information follows the program listing. many thanks in advance Robbie --- Robbie Morrison PhD student -- policy-oriented energy system simulation Institute for Energy Engineering (IET) Technical University of Berlin (TU-Berlin), Germany [from IMAP client] // --------------------------------------------------------- // Copyright: (c) 2007 Robbie Morrison. All rights reserved. // License: Unrestricted reuse in any form, for any purpose. // Version: $Id: frag-xeona-registers-9.cc,v 1.3 2007/10/17 15:23:43 robbie Exp $ // Tested using g++ 4.1.2 (-Wall -Weffc++ -pedantic), boost // 1.34.1 (source build), and valgrind (memory checker) on // Ubuntu 6.10 (7.04 is current) using Linux 2.6.17-12-generic. #include <iostream> // standard io #include <string> // C++ strings #include <vector> // STL sequence container #include <iomanip> // setw() and family #include <tr1/memory> // TR1 smart pointers using std::tr1::shared_ptr; // for convenience // --------------------------------------------------------- template <typename T> class Value // just for type conversion operator T&() { public: explicit Value(T v) : _value(new T(v)) { } operator T&() { return *_value; } // return aliasable value private: shared_ptr<T> _value; // pointer usage is essential }; // --------------------------------------------------------- class Register // initializes and watches selected { // .. Entity variables public: explicit Register() : _vec() { } // explicit for testing template <typename T> // would normally support int,double,bool Value<T> tieRef(const std::string& tag) // would normally search on tag! { T temp = 0.45; // would normally obtain value from user shared_ptr<Value<T> > val(new Value<T>(temp)); // actual object is shared _vec.push_back(val); // crude "last on" indexing for testing report(); return *_vec.back(); // given aforementioned "last on" indexing } void report() // just for doubles (could be templated) { double temp = *_vec.back(); // this line is interesting! std::cout << "Register : value : " << temp << std::endl; } private: std::vector<shared_ptr<Value<double> > > _vec; // data mirror }; // --------------------------------------------------------- class Entity // represents simulation 'things' { public: explicit // explicit for testing Entity(Register& r) : // pass-by-reference necessary _register(r), // for testing purposes _coeff(r.tieRef<double>("coeff")), // intuitive syntax _notInteresting() // rely on default construction { report(); } void modify(double d) { _coeff = d; // normal syntax (no pointer derefs) report(); // display this modification _register.report(); // confirm the modification propagated } void report() { std::cout << "Entity : value : " << _coeff << std::endl; } private: Register& _register; double& _coeff; // watched variable int _notInteresting; // ignored variable }; // --------------------------------------------------------- int main(int argc, char* argv[]) { std::cout << std::fixed << std::setprecision(2); std::cout << " creating a register" << std::endl; Register reg; std::cout << " creating an entity" << std::endl; Entity ent(reg); std::cout << " modifying the entity" << std::endl; ent.modify(0.01); ent.modify(2.0); } // ---- end-of-program ------------------------------------- // OUTPUT // creating a register // creating an entity // Register : value : 0.45 // Entity : value : 0.45 // modifying the entity // Entity : value : 0.01 // Register : value : 0.01 // Entity : value : 2.00 // Register : value : 2.00 // Development environment // // boost : 1.34.1 (via ./configure; make all; make install) // gcc : 4.1.2 20060928 (prerelease) (Ubuntu 4.1.1-13ubuntu5) // os : Ubuntu 6.10, Linux 2.6.17-12-generic (7.04 is current) // hardware : Toshiba Tecra A2 330 laptop (purchased 27-Aug-2004) // specs : 1.4GHz Intel Celeron M (32-bit) / 512MiB RAM / 40GB HDD // // Problem // // The problem involves sharing data between Entity objects, // used to represent various 'things' in my simulation -- // and Register objects ('register' in the sense of a book // recording important information), used to help construct // and then watch the state of Entity objects. A concrete // ObjectFactory object (not relevant to this discussion) // will be used to create the Entity objects (Alexandrescu // 2001 pp179-217). // // I would also like a relatively clear syntax for // programming the Entity's as there are lots of these and // they contain numerical expressions -- for instance, // dereferencing (smart) pointers would yield lines like: // // *_perf = 2.1 * local * *_coeff; // // Solution (provisional) // // The chosen strategy is to declare all to-be-observed // Entity object data members as references and then bind // these to the relevant Register object element upon // construction. // // An alternative approach could be to utilize the Memento // pattern (which relies on friendship) in conjunction with // the Iterator pattern (for traversing a population of // objects) (Gamma etal 1995 pp283-291 and pp257-271 // respectively). // // Design notes // // In relation to C++, a reference data member must be // initialized in the constructor's initializer list // (Lischner 2003 p36). And a reference may not be bound to // a class member (data member or member function) (p35). // // Class Value<T> is instead used to avoid an "invalid // initialization .. from a temporary" error when // initializing Entity::_coeff using the Register::tie<T> // member function. // // Class Value supplies a "type conversion operator" // (Lischner 2003 p112), in this case, Value::operator T&, // which binds to the non-const reference data member in // Entity. The code for Value derives from Lischner (2003 // p119 example 5-21). // // References // // Alexandrescu, Andrei. 2001. Modern C++ design : // generic programming and design patterns applied. // Addison-Wesley, Boston, USA. ISBN 0-201-70431-5. // // Gamma, Erich, Richard Helm, Ralph Johnson, and John // Vlissides. 1995. Design patterns : elements of // reusable object-oriented software. Addison-Wesley, // Boston, USA. ISBN 0-201-63361-2. // // Lischner, Ray. 2003. C++ in a nutshell : a language // and library reference, O'Reilly and Associates, // Sebastopol, California, USA. ISBN 0-596-00298-X.

1 0

Re: [Boost-users] Boost Spirit and C++/CLI
by Hansi Hansi 17 Oct '07

17 Oct '07

When you use windows forms programs you can't use the standard main entry point with mixed unmanaged code. There is a bug in the initialization of the CRT in VS2005. It seems that the bug is not anymore in VS2008. Test it with the entry point '?mainCRTStartupStrArray@@$$FYMHP$01AP$AAVString@System@@@Z' Best regards Hansjörg _______________________________________________________________________ Jetzt neu! Schützen Sie Ihren PC mit McAfee und WEB.DE. 3 Monate kostenlos testen. http://www.pc-sicherheit.web.de/startseite/?mc=022220

3 3

boost::serialization(1.34.1) Can not link 64-bit example.
by Oleg Udovenko 16 Oct '07

16 Oct '07

Hi, Now I'm trying to use the "serialization" library on 64-bit(x64, Visual C++ 8.0). And faced some troubles with that. First I've converted a Visual Studio 2003 solution (vc7ide) to Visual Studio 2005, and compiled 64-bit static library release/debug. It passed the compilation with lots of warnings. But when I tried build a 64-bit example demo_xml_load.cpp I've got a lots of linker errors. - Is somebody tried the "serialization" library on 64-bit? Thanks, Oleg.

1 0

What would happen if boost::shared_ptr is reset to an address in the stack
by Peng Yu 16 Oct '07

16 Oct '07

Hi, The following program runs well in my system. However, if I change 1 to 0, it will terminate at the line with "delete". Does this mean that boost::shared_ptr can tell if the pointer is from the stack or the heap and does different things based on where the pointer is from? Thanks, Peng #include <iostream> #include <boost/shared_ptr.hpp> int main() { #if 1//try change this to 0 int a[100]; boost::shared_ptr<int> p(a); std::cout << std::endl; #else int a[100]; delete a; // would terminate here std::cout << std::endl; #endif }

5 7

[shared_ptr] class ... has pointer data members but does not override ...
by Klaus Backert 16 Oct '07

16 Oct '07

Hello, a Boost newbie question: Arch: PPC, OS: Mac OS X 10.4.10, Compiler: GCC 4.0.1 In a C++ dynamic library, I'm just developing, I have: #include <boost/shared_ptr.hpp> namespace MyNameSpace { typedef boost::shared_ptr<std::string> StringPtr; } I get the following warning: /usr/local/include/boost/shared_ptr.hpp: In instantiation of 'boost::shared_ptr<std::string>': /Volumes/blablabla/MyClass.cpp:81: instantiated from here /usr/local/include/boost/shared_ptr.hpp:131: warning: 'class boost::shared_ptr<std::string>' has pointer data members /usr/local/include/boost/shared_ptr.hpp:131: warning: but does not override 'boost::shared_ptr<std::string>(const boost::shared_ptr<std::string>&)' The incriminated line in MyClass.cpp is the second line here: __attribute__((visibility("default"))) MyNameSpace::StringPtr MyClass::myFunction() const { /* .......... */ } I have experienced the same with classes of my own instead of std::string, i.e. with typedef boost::shared_ptr<MyOtherClass> MyOtherClassPtr; etc. May be it's just me, but the boost documentation and Google don't help. Can I ignore the warnings or should I do something? Thanks in advance Klaus

2 1

How to add a type to is_arithmetic
by Peng Yu 16 Oct '07

16 Oct '07

Hi, Suppose I define a class fixed_point as class fixed_point{ ....some operations are defined here private: int _integer_part; unsigned _fractional_part } I want is_arithmetic<fixed_point>::type to be true_ Of cause, I can partially specialize the template class as template <> struct is_arithmetic<fixed_point> { ... blah blah }; If I use the above approach, shall I put it into the namespace boost or not? Also, I'm wondering if there is any generic way to do this? Thanks, Peng

3 5