December 2010 - Boost-users

[Fusion] Runtime complexity of boost::fusion::find?
by Clinton Mead 10 Dec '10

10 Dec '10

Hi everyone On the following documentation page: http://www.boost.org/doc/libs/1_45_0/libs/fusion/doc/html/fusion/algorithm/… It mentions that the computational complexity for "find" is linear. However, I can not understand how this is the case. As far as I understand all find needs to do is to look through types, simply iterating through the sequence types at compile time using result_of::next<X>::type where where X is initially result_of::begin<SEQ>::type. So I can't see how any actual work is done by find. Am I missing something here? I'm working on a parameter library, and I assumed things like "find" used constant or no time to execute. Could people please explain how find could use linear time? Thanks Clinton

2 1

Re: [Boost-users] variant over base and derived classes
by Hicham Mouline 09 Dec '10

09 Dec '10

> ----Original Message----- >From: "Dave Abrahams" [dave(a)boostpro.com] >Date: 09/12/2010 05:00 PM >To: boost-users(a)lists.boost.org >Subject: Re: [Boost-users] variant over base and derived classes >At Thu, 09 Dec 2010 09:15:45 -0500, >Hicham Mouline wrote: >> >> >> >Subject: Re: [Boost-users] variant over base and derived classes >> >> >(sorry if this is a top post, my phone shows no msg quotes) >> >> >I'm no expert on this, but isn't there an compile time function to >> >testing if an instance is derived from another (in introspection or >> >something)? You could then maybe sort the variant types for most to >> >least derived and then test untill you find the suitable cast? >> >> >Just brainstorming pretty much. Hope its useful! >> >Best, >> >Dee >> >> yes, there is boost::is_base_of or some metafct like that in type_traits. >> not quite sure how to implement the sort though? >> will checkout mpl::sort with binary predicate base_or_derived. >> I have under 10 types, so compile time should be ok >> thanks, >I haven't looked at your use-case in detail, but it's a bit of a >stretch to imagine a case in which runtime polymorphism via virtual >functions would be useful inside a variant, since you need to know the >most derived type of each stored thing anyhow. Are you going to be >referring to the stored items via a base class pointer? >-- >Dave Abrahams >BoostPro Computing >http://www.boostpro.com I have a class hierarchy that I want to be able to mpi::broadcast. mpi::broadcast not being able to polymorphically receive the object on the receiver's end forced me to wrap all the hierarchy in a variant, which I can send and it works at reception as well. at reception, mpi::broadcast requires the real object to exist already. The code that sends with mpi::broadcast receives a base class reference (or pointer). The code that receives with mpi::broadcast doesn't know what it is receiving. Apparently, boost::serialization can deal with base pointers fine, and I think it creates on the heap the loaded objects, or? The solution w variant works here, I am trying to simplify. thanks,

1 0

lib - errata
by Tim St. Clair 09 Dec '10

09 Dec '10

Is there a simple loc to see a libraries errata over versions? -- Cheers, Timothy St. Clair

1 0

Re: [Boost-users] variant over base and derived classes
by Hicham Mouline 09 Dec '10

09 Dec '10

>-----Original Message----- >From: "Diederick C. Niehorster" [dcnieho(a)gmail.com] >Date: 09/12/2010 01:51 PM >To: boost-users(a)lists.boost.org >Subject: Re: [Boost-users] variant over base and derived classes >(sorry if this is a top post, my phone shows no msg quotes) >I'm no expert on this, but isn't there an compile time function to >testing if an instance is derived from another (in introspection or >something)? You could then maybe sort the variant types for most to >least derived and then test untill you find the suitable cast? >Just brainstorming pretty much. Hope its useful! >Best, >Dee yes, there is boost::is_base_of or some metafct like that in type_traits. not quite sure how to implement the sort though? will checkout mpl::sort with binary predicate base_or_derived. I have under 10 types, so compile time should be ok thanks,

2 1

iostream gzip / bzip header
by Kraus Philipp 09 Dec '10

09 Dec '10

Hello, I'm using bzip and gzip iostreams algorithms, but is there a solution to remove the header data of the compressed stream? I need only the compressed data not the header. Thanks Phil

2 1

Re: [Boost-users] variant over base and derived classes
by Hicham Mouline 09 Dec '10

09 Dec '10

-----Original Message----- From: "Hicham Mouline" [hicham(a)mouline.org] Date: 09/12/2010 03:22 PM To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] variant over base and derived classes >>-----Original Message----- >>From: "Diederick C. Niehorster" [dcnieho_at_[hidden]] >>Date: 09/12/2010 01:51 PM >>To: boost-users_at_[hidden] >>Subject: Re: [Boost-users] variant over base and derived classes >>(sorry if this is a top post, my phone shows no msg quotes) >>I'm no expert on this, but isn't there an compile time function to >>testing if an instance is derived from another (in introspection or >>something)? You could then maybe sort the variant types for most to >>least derived and then test untill you find the suitable cast? >>Just brainstorming pretty much. Hope its useful! >>Best, >>Dee >yes, there is boost::is_base_of or some metafct like that in type_traits. >not quite sure how to implement the sort though? >will checkout mpl::sort with binary predicate base_or_derived. >I have under 10 types, so compile time should be ok >thanks, This works. The source is pasted below. I am however obliged to 1. declare a default-constructed variant 2. call mpl::for_each on the sequence of types in the variant a assign functor to assign to the variant. Is there a way I can embed/simplify this 2 stage construction? regards, Here is the source :-) #include <iostream> #include <boost/variant.hpp> #include <boost/type_traits/is_base_and_derived.hpp> #include <boost/mpl/sort.hpp> #include <boost/mpl/vector.hpp> #include <boost/mpl/lambda.hpp> #include <boost/mpl/for_each.hpp> #include <boost/mpl/not.hpp> #include <boost/mpl/logical.hpp> struct S { virtual ~S() {} }; struct Deriv1 : public S { virtual ~Deriv1() {} }; struct Deriv2 : public S { virtual ~Deriv2() {} }; struct Deriv3 : public Deriv2 { virtual ~Deriv3() {} }; typedef boost::variant<S,Deriv1, Deriv2, Deriv3> my_variant_t; typedef boost::mpl::sort<my_variant_t::types, boost::mpl::not_<boost::is_base_and_derived<boost::mpl::_,boost::mpl::_> > >::type v0_sorted_t; template <typename Variant, typename Base> struct assign { assign(Variant& v, const Base& b) : v_(v), assigned_(false), b_(b) {} template <typename T> void operator()(const T&) { std::cout<< typeid(T).name() <<std::endl; if (!assigned_) { const T* t = dynamic_cast<const T*>(&b_); if (t!=0) { v_ = (*t); assigned_ = true; } } } private: Variant& v_; bool assigned_; const Base& b_; }; int main() { S s; Deriv1 d1; Deriv2 d2; Deriv3 d3; const S& base = s; my_variant_t vvv; boost::mpl::for_each<v0_sorted_t>( assign<my_variant_t,S>(vvv, base) ); std::cout<< vvv.which() <<std::endl; return 0; }

2 1

Re: [Boost-users] variant over base and derived classes
by Hicham Mouline 09 Dec '10

09 Dec '10

>>-----Original Message----- >>From: "Diederick C. Niehorster" [dcnieho_at_[hidden]] >>Date: 09/12/2010 01:51 PM >>To: boost-users_at_[hidden] >>Subject: Re: [Boost-users] variant over base and derived classes >>(sorry if this is a top post, my phone shows no msg quotes) >>I'm no expert on this, but isn't there an compile time function to >>testing if an instance is derived from another (in introspection or >>something)? You could then maybe sort the variant types for most to >>least derived and then test untill you find the suitable cast? >>Just brainstorming pretty much. Hope its useful! >>Best, >>Dee >yes, there is boost::is_base_of or some metafct like that in type_traits. >not quite sure how to implement the sort though? >will checkout mpl::sort with binary predicate base_or_derived. >I have under 10 types, so compile time should be ok >thanks, This works. The source is pasted below. I am however obliged to 1. declare a default-constructed variant 2. call mpl::for_each on the sequence of types in the variant a assign functor to assign to the variant. Is there a way I can embed/simplify this 2 stage construction? regards,

1 0

Re: [Boost-users] mpi/serialization: broadcast function and the value argument
by Hicham Mouline 09 Dec '10

09 Dec '10

-----Original Message----- From: "Matthias Troyer" [troyer(a)phys.ethz.ch] Date: 09/12/2010 02:39 PM To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] mpi/serialization: broadcast function and the value argument >On Dec 8, 2010, at 5:50 PM, Hicham Mouline wrote: >> Following a suggestion from another thread, I have tried broadcasting in a polymorphic way. >> >> I have a hierarchy of polymorphic struct (they have 2 virtual functions). I was wondering if it was possible/allowed to >> >> root process: >> base* b; /// b points to an actual derived struct but may also point the base struct (not abstract) >> broadcast( ..., b, ...); >> >> slave processes: >> base* b; >> broadcast( ..., b, ...); /// broadcast constructs the appropriate struct and makes b point to where it did that >> >> assuming all the conditions for pointer to class conditions are satisfied for the serialization library. >> >If serialization works on your polymorphic class then this should work as well. >Matthias It appears the implementation of mpi::broadcast(), according to this post, http://lists.boost.org/boost-users/2010/12/64802.php and I confirm that expects an already built object to call the (de)serialization procedure on. Therefore, this seems a specific issue that mpi::broadcast() may want to handle differently, could you confirm/correct? thanks,

1 0

serialization: const char*
by Hicham Mouline 09 Dec '10

09 Dec '10

hello, serializing a const char* is not allowed I gather? char* seems not to be allowed either. primitive types pointers are not serializable? I have a struct S{ double d; char c[SIZE]; const char* x; }; template <typename Archive> void serialize(Archive& ar, S& s, const unsigned int version) { ar & s.d; ar & s.c; ar & s.x; } the 3rd one seems to cause the compile error, removing constness didn't help. is one supposed to an array of char or a std::string instead regards,

3 2

mpi/serialization: broadcast function and the value argument
by Hicham Mouline 09 Dec '10

09 Dec '10

Following a suggestion from another thread, I have tried broadcasting in a polymorphic way. I have a hierarchy of polymorphic struct (they have 2 virtual functions). I was wondering if it was possible/allowed to root process: base* b; /// b points to an actual derived struct but may also point the base struct (not abstract) broadcast( ..., b, ...); slave processes: base* b; broadcast( ..., b, ...); /// broadcast constructs the appropriate struct and makes b point to where it did that assuming all the conditions for pointer to class conditions are satisfied for the serialization library. Otherwise, perhaps a solution with a boost::variant over the base struct and all of its derived is the way to go? Thanks,

2 1