September 2002 - Boost-users

RE: [Boost-Users] Re: Use a shared_ptr as a 'backpointer'?
by Mark Storer 30 Sep '02

30 Sep '02

I've got a comment and a question... 1) Couldn't the "backpointer" just be a constant reference? Backpointers, in my mind, don't have anything to do with ownership, so a shared_ptr woulnd't be appropriate. 2) shared_from_this? I can't find that one in any of the headers or docs I downloaded with 1.28. I did find shared_count in (oddly enough) shared_count.hpp, which looks like it would prevent two seperate shared_ptr's from causing a double-delete... pretty slick, but I couldn't find any docs on it. --Mark Storer Software Engineer Cardiff Software #include <disclaimer> typedef std::disclaimer<Cardiff> Discard; -----Original Message----- From: David Abrahams [mailto:yg-boost-users@m.gmane.org] Sent: Sunday, September 29, 2002 4:12 PM To: boost-users(a)yahoogroups.com Subject: [Boost-Users] Re: Use a shared_ptr as a 'backpointer'? <dick.bridges(a)tais.com> wrote in message news:OF699D7956.EED22A4F-ON88256C43.00722AD0@tais.net... > I run into situations where it's useful for an object to have a > 'backpointer' to the object instance that created it (example below). Is > there a pattern/idiom for doing this using shared_ptr? Or maybe I need to > re-think the idea of the backpointer in the first place? > class Bar; class Foo { public: Foo(shared_ptr<Bar>); private: shared_ptr<Bar> p; }; class Bar : boost::shared_count { void doit() { new Foo(boost::shared_from_this(this)); // <<< new Foo(this) obviously won't work. } }; HTH, Dave -- ----------------------------------------------------------- David Abrahams * Boost Consulting dave(a)boost-consulting.com * http://www.boost-consulting.com Info: <http://www.boost.org> Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

1 0

RE: [Boost-Users] shared_ptr and counted_base
by Hickman, Greg 30 Sep '02

30 Sep '02

> I was looking through the latest boost documentation, trying Where is the documentation to which you refer? Is this a development doc or part of the ordinary user docs (if the latter, I don't know where to find it)? -Greg

2 1

Problem using 'shared_ptr<T>::operator impl-defined-type()'
by Hickman, Greg 30 Sep '02

30 Sep '02

With Borland 5.5.1, I'm not having any luck using a shared_ptr in a boolean context. The compiler complains that the 'if' condition below contains an "Illegal structure operation". Is this a known deficiency with my compiler? -Greg Hickman #include <boost/shared_ptr.hpp> int main(int argc, char* argv[]) { boost::shared_ptr<int> intp; if (intp) // error here ; return 0; }

5 10

RE: [Boost-Users] Returning a boost::scoped_lock by value
by Darryl Green 30 Sep '02

30 Sep '02

This issue has come up before on the development list (search for Boost.Threads Locking Delima). I'm not sure that it was resolved satisfactorily - move semantics for scoped_locks seemed to be the answer, but I think there were problems actually implementing that. The intent was that if move couldn't be implemented, then the mutex lock methods would have to be exposed. > -----Original Message----- > From: Raoul Gough [mailto:yg-boost-users@m.gmane.org] > Sent: Friday, 27 September 2002 9:44 PM > To: boost-users(a)yahoogroups.com > Subject: [Boost-Users] Returning a boost::scoped_lock by value > > > > boost::scoped_lock is non-copyable, so it is impossible to > return one from a function by value. However, I would like to > be able to do something like this: > > mt::holder<std::ostream> logStream; > > logStream() << "Hello world\n"; > > where the operator() returns a proxy object that contains a > scoped_lock and a pointer to the data (the ostream). What I > currently have is this: > > (mt::proxy<std::ostream> (logStream)) << "Hello world\n"; > > So the client code has to know what type the proxy object is > and construct one itself. If scoped_lock would provide access > to its associated mutex object, this wouldn't be a necessary, > because the proxy object could then have a copy-constructor, > and operator() in the holder template would look like this: > > proxy<T> operator() () > { > proxy<T> temp (*this); > return temp; > } > > The proxy copy constructor would unlock() the original lock > and construct the copy using the original's mutex (which is > the part that can't currently work, because there is no > public access to a lock's mutex). Assuming that the compiler > implements the NRVO, the unlock/lock sequence wouldn't > actually happen, and the client code just gets its proxy > object constructed for it. If the compiler doesn't implement > the NRVO, the code still works, but is less efficient. > > I guess exposing the raw mutex from scoped_lock would break > an important level of encapsulation. Still, it would be kind > of nice in this case. Maybe if proxy<T> were part of the > library and made a friend of scoped_lock? The only other > alternative I've come up with is to allocate the lock on the > heap and return an object containing a std::auto_ptr (which > makes usage easier, but less efficient). Any comments or suggestions? > > Regards, > Raoul Gough. > > > // Returnable proxy template > > template<typename T> > class proxy > { > boost::mutex::scoped_lock mLock; > T *mPtr; > > boost::mutex &release() > { > mLock.unlock(); > return mLock.mutex(); // <<<< doesn't work > } > > public: > > proxy (proxy &other) > : mLock (other.release()) > , mPtr (other.mPtr) > { > } > > // ... > }; > > > > > > > ------------------------ Yahoo! Groups Sponsor > ---------------------~--> Home Selling? Try Us! > http://us.click.yahoo.com/QrPZMC/iTmEAA/MVfIAA> /EbFolB/TM > > > -------------------------------------------------------------- > -------~-> > > Info: <http://www.boost.org> > Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> > Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> > > > Your use of Yahoo! Groups is subject to > http://docs.yahoo.com/info/terms/ > >

2 1

RE: [Boost-Users] newbie Question about the macro BOOST_JOIN
by scleary＠jerviswebb.com 28 Sep '02

28 Sep '02

> From: zhangwusheng [mailto:yg-boost-users@m.gmane.org] > > Hi,all, > In file suffix.hpp there are lines: > > #define BOOST_JOIN( X, Y ) BOOST_DO_JOIN( X, Y ) > #define BOOST_DO_JOIN( X, Y ) BOOST_DO_JOIN2(X,Y) > #define BOOST_DO_JOIN2( X, Y ) X##Y > > > why donot we define such macro instead: > #define BOOST_JOIN( X, Y ) X##Y The idea behind BOOST_JOIN is to force the preprocessor to scan the macro parameters for replacement before merging the results with ##. So, it changes the meaning if one of the parameters is a macro: #define Z_BOOST_JOIN(X, Y) X##Y Z_BOOST_JOIN(id_, __LINE__) // Expands to "id___LINE__" BOOST_JOIN(id_, __LINE__) // Expands to "id_13" if __LINE__ is 13 However, I'm not sure why BOOST_DO_JOIN2 is used; it seems to me that only one layer of indirection should work. I assume it's to work around some compiler's non-conforming preprocessor. -Steve

3 2

Boost::UBLAS problems
by pecos_outwest 27 Sep '02

27 Sep '02

Hi Folks, I have been experimenting with the new BOOST linear algebra section boost/numeric/ublas I have been running the examples and benchmark programs bench1 and test1 provided with the distribution. Both programs compile and link perfectly they also run to completion; but they only run one time ! I cannot run a second time. It looks like the executable isn't even starting. Has anyone found this problem and found the solution I am using Microsoft VC6 compiler on a PC Thanks...Pecos

1 0

shared_ptr and counted_base
by Bobby Thomale 27 Sep '02

27 Sep '02

I was looking through the latest boost documentation, trying to find a way to get a shared_ptr from a this pointer of an object that already is contained by a shared_ptr somewhere, without knowing about that other shared_ptr out there. I noticed this intriguing comment in the latest docs: > [This constructor has been changed to a template in order to remember the > actual pointer type passed. The destructor will call delete with the same > pointer, complete with its original type, even when T does not have a virtual > destructor, or is void. > > In the current implementation, if p is convertible to counted_base *, > shared_ptr will use the embedded reference count supplied by counted_base. > This is an (experimental) attempt to provide a way for shared_ptr to be > constructed from a raw pointer such as this. A free function > shared_from_this(q) performs the conversion when q is convertible to > counted_base const *. > > The optional intrusive counting employed by the current implementation allows > shared_ptr to interoperate with intrusive_ptr, an experimental generic > intrusive-counted smart pointer. > > Another possible implementation is to use a global pointer-to-count map > instead of intrusive counting. shared_from_this would no longer be O(1), which > is a concern for some users, although I do not expect any performance > problems, since the operation is rare. Maintaining a global map is difficult; > it needs to be initialized before any shared_ptr instances are constructed, > and the initialization needs to be thread safe. In addition, under the Windows > dynamic library model, it is possible for several maps to exist. > > It is not yet clear which implementation should be used, or whether the > specification should allow both; nevertheless, the ability to make a > shared_ptr from this is considered essential by experienced smart pointer > users.] That looks like EXACTLY what I am looking for. But a couple things aren't clear to me: 1. Usage - The docs tell me its there but don't really say explicitly how to use it. It looks like, to use this, all I have to do is subclass boost::counted_base in my class that is being shared. Then I can just call shared_from_this to get another shared_ptr out of myself and it should just work. If that is all I have to do, then awesome! Is it really that easy or is there another step I am missing? 2. The word "experimental" above scares me. :-) If I use the counted_base subclass and shared_from_this function, is that code going to break later on if you guys change your minds? Anyway, I just want to reiterate what the comment at the end says, that the ability to make a shared_ptr from *this is essential to use these things. :-) -- Bobby --------------------------------------------------------------------- Bobby Thomale Senior Software Developer Inoveon Corporation http://www.inoveon.com/ ---------------------------------------------------------------------

2 2

Returning a boost::scoped_lock by value
by Raoul Gough 27 Sep '02

27 Sep '02

boost::scoped_lock is non-copyable, so it is impossible to return one from a function by value. However, I would like to be able to do something like this: mt::holder<std::ostream> logStream; logStream() << "Hello world\n"; where the operator() returns a proxy object that contains a scoped_lock and a pointer to the data (the ostream). What I currently have is this: (mt::proxy<std::ostream> (logStream)) << "Hello world\n"; So the client code has to know what type the proxy object is and construct one itself. If scoped_lock would provide access to its associated mutex object, this wouldn't be a necessary, because the proxy object could then have a copy-constructor, and operator() in the holder template would look like this: proxy<T> operator() () { proxy<T> temp (*this); return temp; } The proxy copy constructor would unlock() the original lock and construct the copy using the original's mutex (which is the part that can't currently work, because there is no public access to a lock's mutex). Assuming that the compiler implements the NRVO, the unlock/lock sequence wouldn't actually happen, and the client code just gets its proxy object constructed for it. If the compiler doesn't implement the NRVO, the code still works, but is less efficient. I guess exposing the raw mutex from scoped_lock would break an important level of encapsulation. Still, it would be kind of nice in this case. Maybe if proxy<T> were part of the library and made a friend of scoped_lock? The only other alternative I've come up with is to allocate the lock on the heap and return an object containing a std::auto_ptr (which makes usage easier, but less efficient). Any comments or suggestions? Regards, Raoul Gough. // Returnable proxy template template<typename T> class proxy { boost::mutex::scoped_lock mLock; T *mPtr; boost::mutex &release() { mLock.unlock(); return mLock.mutex(); // <<<< doesn't work } public: proxy (proxy &other) : mLock (other.release()) , mPtr (other.mPtr) { } // ... };

1 0

[BGL]: Iterating through internal properties of specific value
by Björn Lindberg 27 Sep '02

27 Sep '02

I have a BGL graph (adjacency_list). I use it to represent a tree, and so I have an internal vertex property of type enum representing the node type, ie ROOT, LEAF, INNER. Now, I would like a way to access only the leaves of a tree, preferrably through iterators. I can't think of any good way to accomplish this, except possibly by writing new iterator classes which would internally keep track of a property map and thus hide traversal of the non-leaf nodes from the code using the iterators. This seems like a cumbersome approach though, and I'm hoping there is a simpler way? Björn

3 5

RE: [Boost-Users] metaprogramming and MPL
by Aleksey Gurtovoy 27 Sep '02

27 Sep '02

Ihsan Ali Al Darhi wrote: > I am interested in metaprogramming and Boost mpl. But I can't find a > complete doc about the two subjects together. Where on the > net I can such info? It's not exactly a "complete doc" yet, but it's the most of that is publicly available to the moment - http://www.mywikinet.com/mpl/. Note that MPL sources in the CVS main trunk are more up-to-date than the .zip archive available for download from the library's page. HTH, Aleksey

1 0