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

Re: [Boost-users] downcasting shared pointer
by Bjorn.Karlsson＠readsoft.com 31 Oct '05

31 Oct '05

Hello Faheem, > From: Faheem Mitha [mailto:faheem@email.unc.edu] > > It would be nice if there was some method to downcast the shared_ptr > itself. See http://www.boost.org/libs/smart_ptr/shared_ptr.htm#static_pointer_cast for details on how to do just that. Regards, Bjorn Karlsson

2 1

Implementing lock/unlock with boost::thread
by Johan Torp 31 Oct '05

31 Oct '05

I'm trying to implement a class having a lock and an unlock-method. If several different threads call scoped_lock::lock, what is the effect? class AntiRAII{ public: AntiRAII() : mLock(mMutex, false) {} void Lock() {mLock.lock(); } void Unlock() {mLock.unlock();} private: boost::mutex mMutex; boost::mutex::scoped_lock mLock; }; I cannot find any documentation on the behaviour. The usual usage of scoped_lock is one per thread, which is why I'm sceptical to this approach. How can this be implemented? Is it better to expose the mutex? Regards, Johan Torp

3 2

Re: [Boost-users] [Multi-Index] composite_key problems
by JOAQUIN LOPEZ MU?Z 31 Oct '05

31 Oct '05

----- Mensaje original ----- De: Andreas Volz <lists(a)brachttal.net> Fecha: Domingo, Octubre 30, 2005 8:15 pm Asunto: [Boost-users] [Multi-Index] composite_key problems > Hi, > > I like to build a data structure and like to access it with different > keys, so I try multi_index for this. [...] > I hacked the basic example to get this working. My example compiles, > but if I execute it, I get only warnings: > > probe: /usr/include/boost/multi_index/detail/index_iterator.hpp:107: > typename Node::value_type& > boost::multi_index::detail::index_iterator<Node, > Container>::operator*() const [with Node = > Container>boost::multi_index::detail::ordered_index_node<boost:: > multi_index::detail::index_node_base<learnedMeasure> > Container>>, Container = > Container>>boost::multi_index::detail::ordered_index<boost:: > multi_index::composite_key<learnedMeasure, > Container>>boost::multi_index::member<learnedMeasure, std::string, > Container>>&learnedMeasure::measu > .... more warnings > > I put my example code online. Perhaps someone could take a look in it > and help me to find this problem. > > http://cpp.sourceforge.net/?show=9550 > Hello Andreas, I can't compile your source right now, but from the looks of it I cannot see anything wrong. What compiler are you using? If MSVC++ 6.0 or 7.0, take into account they have non-conformances you must take into account when using Boost.MultiIndex, see http://boost.org/libs/multi_index/doc/compiler_specifics.html Your particular code won't work on either of these compilers due to problems with member<> and composite_key that are addresses in the aforementioned page. Is this the problem? If not, I'll try to get back to you in a couple of days. Best, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

regexp: New problem with trailing - in character class
by Greg Ferrar 31 Oct '05

31 Oct '05

Hi, I've been using Boost regexp library forever, and it's great. But I just grabbed 1.33.0 (trying to get it to compile on 64-bit Windows), and I'm getting an error I never got before: it refuses to allow this regular expression: [a-z-] which is supposed to mean "a lower case letter or a dash". Previous versions of Boost regexp allowed this, and grep allows it, and version 1.33.0 does allow this one: [-a-z] Is this a deliberate restriction? It's causing me lots of problems, because my application has a lot of existing regular expressions which use this syntax. This occurs both on x64 with VC8, and on a pretty standard (and old) 32-bit Linux system. I've hacked around it horribly for now by adding this to the top of regcompA (I use the A posix interface exclusively): // HACK by GMF to fix problem that [a-z-] is not accepted as valid, but [-a-z] is. Fix it by moving the dash. char *p = (char*) ptr; while (*p) { printf("p: %c\n", *p); if (*p == '\\') p++; else if (*p == '[') { char *q = p+1; while (*q && (*q != ']')) { printf("*q: %c\n", *q); q++; } if (*q == ']') { q--; if (*q == '-') { memmove(p+2, p+1, (q-p)-1); *(p+1) = '-'; p = q; } } } p++; } but that's very nasty, and probably doesn't work properly anyway, and I'd sure like to get that out of my production code. Help! Greg

2 1

Using iterator_facade from base class with sub classes
by Daniel Lidström 31 Oct '05

31 Oct '05

Hello! I have implemented the iterator_facade for a base class called DataObject. So now I am able to write things like this: boost::shared_ptr<PxyFileData> file(...); const DataObject& points = file->GetPoints(); int num_points = std::distance(points.begin(), points.end()); //! random access iterator A DataObject acts like a tree structure. Each DataObject holds a vector of pointers to children, which are also DataObjects. Now to my problem: A PxyPoint is a sub class to DataObject, and PxyFileData::GetPoints() returns a DataObject where all the children are PxyPoint's. Now I would like to call a member function, PxyPoint::PutSpCode(const char*), on each of the children, like this: void put_sp_code(DataObject&, const char*); void test() { shared_ptr<PxyFileData> file(...); DataObject& points = file->GetPoints(); put_sp_code(points, "spcode"); } void put_sp_code(DataObject& points, const char* code) { std::for_each(points.begin(), points.end(), std::bind2nd(std::mem_fun(&PxyPoint::PutSpCode), code)); } This does not compile (VC 7.1): c:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\algorithm(21): error C2664: 'binder2nd<_Fn2>::result_type binder2nd<_Fn2>::operator ()(binder2nd<_Fn2>::argument_type & ) const': cannot convert parameter 1 from 'DataObject' to 'binder2nd<_Fn2>::argument_type & ' with [ _Fn2=mem_fun1_t<void,TPunkt,const char *> ] and [ _Fn2=mem_fun1_t<void,TPunkt,const char *> ] A reference that is not to 'const' cannot be bound to a non-lvalue c:\projects\GEOW\V12\Test\GeoTest\Model Test\ModelTest.cpp(48): see reference to function template instantiation '_Fn1 for_each<iter,binder2nd<_Fn2> >(_InIt,_InIt,_Fn1)' being compiled with [ _Fn1=binder2nd<mem_fun1_t<void,TPunkt,const char *> >, _Fn2=mem_fun1_t<void,TPunkt,const char *>, _InIt=iter ] (error message formatted with STLTask) Is this possible with our design? I would like to be able to do this without having to create a PxyPointList, that derives from DataObject and only stores PxyPoint's (my guess is that something like that would work with this code). Thanks in advance! Regards, Daniel

3 2

Re: [Boost-users] [multi_index] non-intrusive keys
by JOAQUIN LOPEZ MU?Z 31 Oct '05

31 Oct '05

----- Mensaje original ----- De: Adi Shavit <adish(a)gentech.co.il> Fecha: Viernes, Octubre 28, 2005 4:11 pm Asunto: [Boost-users] [multi_index] non-intrusive keys > Hi, > > In my application I used a multi_map of multi_maps as a data > structure to provide indexing with 2 keys, say long strings. Access > is > logarithmic on both keys and the keys are not part of the stored > object, > thus are potentially only stored once. > > I'd like to use multi_index as a more elegant replacement. > However, I'm not clear about where the keys are stored. > > In my app, the keys are stored separately from the data, as the > keys to > maps, as part of the map data structure. > It seems that in multi_index the keys have to be inside the stored > data.In this case, 10 objects with the same key(s) will all have to > hold > copies of the keys (as members). > > Is there a way to avoid this? > I don't think you can avoid duplication of keys --if the keys are really expensive to duplicate, you might consider holding shared_ptrs to them, but this is AFAICS the furthest you can get. Take into acount that a multi_index_container with two keys is not a map of maps. These are different beasts, serving different purposes, * a map of maps applies two concatenated lookups to get to a "terminal" object, * a multi-index container with two indices applies either one of two poosible lookups to get to the object, so you might want to ask yourself whether it makes sense to try to replace your current code with Boost.MultiIndex. What are the advantages / use cases you envision? Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

Problems comping Boost.Iostreams on Windows with zlib support
by Adam M 31 Oct '05

31 Oct '05

Hi all, I am trying to compile Boost on my Windows 2000 box, but I need to enable the zlib support in Boost.Iostreams for the project I'm working on. I read through the "Getting Started" guide and the Boost.Iostreams docs, and I have done what it says to have zlib support installed, but it's not working. I downloaded the zlib DLL from their website and extracted it to C:/zlib Here's my bjam command line: bjam "-sNO_COMPRESSION=0 -sTOOLS=vc-7_1 -sNO_ZLIB=0 -sZLIB_BINARY=zdll -sZLIB_INCLUDE=C:/zlib/include -sZLIB_LIBPATH=C:/zlib/lib" install When I do that, bjam says: "Building Boost.Iostreams with zlib and gzip support disabled. To enable zlib and gzip, consult the Boost.Iostreams documentation." I thought maybe it was a red herring, so I let the compile finish anyway. However, there's no "libboost_zlib-*" in my Boost/lib folder. Can someone enlighten me as to what I've done wrong here? I'm stumped. Thanks, Adam

2 1

Templates Question
by Surya Kiran Gullapalli 31 Oct '05

31 Oct '05

Hi, I'm writing a templated class, and i dont want to use the class otherthan for some predetermined types, say, int, double etc. This class has no meaning for typenames other than those few. ========================= template <class T> myClass {....} ; myClass<int> intClass ; myClass<float> floatClass ; myClass<char> charClass ; ========================= The compiler should not complain for the first 2 instantiations, but should bail out at the 3rd instantiation (charClass). Is there any way i can achieve this, using standard c++ or boost. I've tried explicit template instantiation, (by passing a flag to my compiler), but in that case, i've to do that for every class suppose i try to invoke std::vector <someOtherClass> someVector; I've to instantiate this also explicitly. (which is a problem for me, because, I use a lot of standard library, and its a pain for me to instantiate every time. Also, if i do the above, i can always instantiate myClass also like this. template class myClass <someOtherClass> ; and which i want to avoid this.. to summarize my question, is there any way i can restrict the compiler to accept only few type names for my templated class (i.e., using standard c++ or boost) Thanks, Surya

4 5

Using a set with kruskal_minimum_spanning_tree
by Jeffrey Holle 30 Oct '05

30 Oct '05

I've sucessfully created an program that uses kruskal_minimum_spanning_tree. I presently give a vector container to this algorithm, which it populates in providing its answer. Now I want to optimize my program by switching to set. For vector, I use "back_inserter(m_spanningTreeEdges)". For set, I've attempted to use "inserter(m_spanningTreeEdges,m_spanningTreeEdges.end())", but this doesn't work. Can somebody tell me the correct way to code this, or tell me that it isn't possible?

1 1

downcasting shared pointer
by Faheem Mitha 30 Oct '05

30 Oct '05

Hi, I was wondering what the recommended way was to downcast shared pointers. I have a a base class, Base, and a derived class Derived. I have a Base shared pointer which points to a Derived object. shared_ptr<Base> d; As part of a unit test, I'm trying to extract the Derived object from it. *static_cast<Derived *>(d.get()); Here I'm extracting the raw pointer, downcasting it to Derived, and then taking the value. It would be nice if there was some method to downcast the shared_ptr itself. Thanks in advance. Please CC me on any reply. Faheem.

1 0