- Boost-users - lists.boost.org

Another threads question...
by Martin Johnson 19 Dec '02

19 Dec '02

Hi, Just to get it through my thick skull. >The threadmon.dll is a seperate library that's always dynamically linked. Then why not make threads a dynamic lib. As an exercise I attempted this though as I found that the template methods of condition a little troublesome. Could this not be reworked by reworking or typing the condition class. I am able to build the lib as a dll with threadmon.cpp included. It involves exporting all classes and crude hack to condition *just* to get it to work. It would appear to function correctly as far as I can see. With the cleanup handlers being called without requiring a seperate dll, though I guess this puts a static link a little out of the picture. But why would I static link anyway, for distribution purposes, but I have to distribute threadmon.dll What do you think? I am just trying to think of a way around it, I know you have been there done that but i just need to appreciate this a little better. Regards Martin Johnson

2 1

MacOS X MSL Mach-O Problem with Threads
by Bobby Thomale 19 Dec '02

19 Dec '02

It's me again with weird problems compiling Boost on the Mac! :-) I just converted my MacOS X code from CFM to Mach-O and I ran into another small compilation problem. This time, it is the thread code. Now that I am moved over to Mach-O, the _POSIX_THREADS flag is true, causing posix_features.hpp to set BOOST_HAS_PTHREADS to true. This seems to be causing all sorts of compiler errors. Probably due to the fact that, even though some of the posix headers are being included, I am using the MSL headers rather than the BSD headers, and my understanding is that none of that posix stuff will actually work until I eject the MSL in favor of the Unix C headers, which I am not prepared to do right now. That and the fact that since I am using Carbon, TARGET_CARBON is also true, therefore BOOST_HAS_MPTASKS is getting set to true. So anyway ... BOOST_HAS_PTHREADS and BOOST_HAS_MPTASKS are both getting defined. I would like to be able to control whether BOOST_HAS_PTHREADS gets defined, because I am actually wanting to use the MP libraries instead, despite _POSIX_THREADS being defined. It looks like how you handled this same problem with Windows was by adding a !defined(BOOST_HAS_WINTHREADS) to the condition as follows: # if defined(_POSIX_THREADS) && (_POSIX_THREADS+0 >= 0) && !defined(BOOST_HAS_WINTHREADS) # define BOOST_HAS_PTHREADS # endif One solution would be to add an additional check for MP Tasks as follows: # if defined(_POSIX_THREADS) && (_POSIX_THREADS+0 >= 0) && !defined(BOOST_HAS_WINTHREADS) && !defined(BOOST_HAS_MPTASKS) # define BOOST_HAS_PTHREADS # endif I did this in my local working copy and it seems to have fixed the problem. I had to add an additional "#define BOOST_HAS_MPTASKS" to my user.hpp to make sure that BOOST_HAS_MPTASKS got defined before the above check. I am not sure if this is the best way or if there is a better way to solve this problem. I suppose another approach would be to check for !defined(TARGET_CARBON) in the above check - if carbon is on, MP Tasks will get turned on. Or you could in the place where BOOST_HAS_MPTASKS gets defined, undef the BOOST_HAS_PTHREADS. But the way I solved it in my local copy (shown above) using !defined(BOOST_HAS_MPTASKS) seemed the cleanest to me, especially since it looks like it is how you solved the same problem for Windows. Anyway - I tried hard to find a way to make this compile without changing anything but my user.hpp but I failed. (If someone has any ideas I'd love to hear them.) -- Bobby --------------------------------------------------------------------- Bobby Thomale Senior Software Developer Inoveon Corporation http://www.inoveon.com/ ---------------------------------------------------------------------

2 2

compilation error with boost_1.29.0
by Volker Boerchers 19 Dec '02

19 Dec '02

Hi, I'm getting the following error on Linux 2.2.16 / Suse 7.0 / g++ 2.95.2 / glibc 2.1.3: ../../../../../include/boost/pool/detail/ct_gcd_lcm.hpp:43: sorry, not implemented: `static_cast_expr' not supported by dump_expr ../../../../../include/boost/pool/detail/ct_gcd_lcm.hpp:43: sorry, not implemented: `static_cast_expr' not supported by dump_expr The following patch fixes it: --- include/boost/pool/detail/ct_gcd_lcm.hpp.orig Thu Dec 19 15:10:54 2002 +++ include/boost/pool/detail/ct_gcd_lcm.hpp Thu Dec 19 14:58:41 2002 @@ -38,7 +38,7 @@ BOOST_STATIC_CONSTANT(unsigned, A_mod_B_ = A % B); BOOST_STATIC_CONSTANT(unsigned, value = (::boost::details::pool::details::ct_gcd_helper< - B, static_cast<unsigned>(A_mod_B_), + B, A_mod_B_, ::boost::type_traits::ice_eq<A_mod_B_, 0>::value >::value) ); }; What is the cast good for anyway? It seems redundant. Regards, Volker -- Volker Boerchers <vboerchers(a)tecon.de> System Engineer TECON Systems AG, Perlengraben 2, 50676 Köln, http://www.tecon.de Tel: +49-221-92007-55, Fax: +49-221-92007-77

1 0

Preprocessor library question
by Tanton Gibbs 19 Dec '02

19 Dec '02

I'm trying to create a template class based on user input. For example: template< int i, int j, int k > class T {}; int main() { int i, j, k; cin >> i >> j >> k; T<i, j, k> t; } Naturally, this is not possible, as i, j, and k must be constants. Therefore, I would like to limit the maximum value of i, j, and k to 100 and create all permutations for T<i,,j,k> at compile time. Basically, something like this: int main() { int i, j, k; cin >> i >> j >> k; if( i == 0 && j == 0 && k == 0 ) { T<0,0,0> t; } else if( i == 0 && j == 0 && k == 1 ) { T<0,0,1> t; } ... else if( i == 100 && j == 100 && k == 100 ) { T<100,100,100> t; } } Naturally, I don't want to type all that in myself, so I was hoping the boost preprocessor library could do it for me. If someone knows of a way to do it, please let me know. Thanks! Tanton [Non-text portions of this message have been removed]

2 1

Matrix-Matrix reference
by Wilken, Marc 18 Dec '02

18 Dec '02

Hello, does ublas support a matrix method like blitz++ .reference() method? (This method allows different matrices to share the same matrix datas.) Marc Wilken This e-mail contains confidential information for the exclusive attention of the intended addressee. Any access of third parties to this e-mail is unauthorised. Any use of this e-mail by unintended recipients such as copying, distribution, disclosure etc. is prohibited and may be unlawful. When addressed to our clients the content of this e-mail is subject to the General Terms and Conditions of GL's Group of Companies applicable at the date of this e-mail. GL's Group of Companies does not warrant and/or guarantee that this message at the moment of receipt is authentic, correct and its communication free of errors, interruption etc.

2 2

Problem with "this" and shared_ptr
by Michael Brinkmann 18 Dec '02

18 Dec '02

Hello, given the following code which uses shared_ptrs, I'm getting a double-free onto an object of type A: ------------------------------------------------------ #include <boost/shared_ptr.hpp> #include <list> #include <iostream> class A { public: A(void){} void makeSomething(void); void makeSomethingOther(void) { std::cout << "A" << std::endl; } }; typedef boost::shared_ptr<A> APtr; class B { public: B(const APtr& a) : mA(a) {} void makeSomething(void) { std::cout << "B" << std::endl; mA->makeSomethingOther(); } private: APtr mA; }; void A::makeSomething(void) { B b(this); // here I want to hand over the "A" shared pointer to "B", but I have only my "this" at hand. std::cout << "A" << std::endl; b.makeSomething(); } int main(void) { APtr a(new A); a->makeSomething(); } ------------------------------------------------------ (please don't ask me about why I'm doing it in that way, the code is stripped down from a much bigger program) I understand that the problem results from the accesss to "this" in the commented line. Is there an elegant way to get the shared_ptr pointing to "this" from "this", which would be the solution to the fault? Or do I have to store a shared_ptr member variable referencing "this" in the class A which is set by a A::method from outside of A to reference it? (not very beautiful) Sorry if this is a FAQ, but I didn't find a solution on news.gmane.org kind regards Michael

4 6

Re: [Boost-Users] Re: Thread library: get thread id
by dick.bridges＠tais.com 18 Dec '02

18 Dec '02

<alternative exemplar> When I implemented the Leader-Follower thread pattern described by Schmidt, et al in Pattern-Oriented Software Architecture, I couldn't figure out a way around comparing the thread id to the current leader in the promote_new_leader function. Probably lack of imagination and/or experience on my part, but having the thread id sure made it easier. </alternative exemplar> "William E. Kempf" To: <Boost-Users(a)yahoogroups.com> <wekempf(a)cox.net cc: > Subject: Re: [Boost-Users] Re: Thread library: get thread id 12/16/2002 01:20 PM Please respond to Boost-Users olivier_debels said: > --- In Boost-Users(a)yahoogroups.com, "William E. Kempf" <wekempf@c...> > wrote: >> >> olivier_debels said: >> > >> > I'm using boost_1_29_0 and need to get the id of the current > thread. >> >> Why? >> > I have several threads which pop messages from the queue. I want to > keep track of what each thread pops (depending on some rules) without > adding an extra parameter (threadId) in the pop-function. I don't follow this. > By this way threads just call the pop-function (they don't know > anything about the internal cooking of the queue). They get a > message, which fulfills the rules (one of the rules is f.e. that no > two threads can handle the same message at the same time). How does a thread id figure into this? What little of this description I understand gives no need for a thread id. William E. Kempf 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/

3 4

switching to BGL
by Daniel Varga 18 Dec '02

18 Dec '02

Hi, I'm planning to throw away my own little undirected graph implementation in favor of Boost Graphs. My implementation is based on an STL map< Node,set<Node> > adjecency map. I see that it's very easy to get the same functionality using the BGL adjecency_list. But what I really like about my implementation is that I have const set<Node>& Graph::lookupNeighbours(const Node&) const; This means that in no time I can get a _container_ of the neighbours, and not just a pair of iterators. Of course teoretically they are equivalent. But I want to rewrite as little code as possible, so I would like to wrap my old graph interface around the Boost Graph. (Not hiding its own, better interface, of course. :) ) Should I write a wrapper container class simulating const set<Node>? Or does adjecency_list expose the very same const set<Node>& for me?

1 0

Threads question
by Martin Johnson 18 Dec '02

18 Dec '02

Hi, This has probably been asked before. I cant find doco giving detail. So excuse me if this is old topic. I am involved with GPL project that uses boost::threads. I have a problem understanding implementation and what threadmon.dll is for. I see that the code is attempting to manage tss stuff. How is the tss stuff managed when lib is a static link under win32? Is it correct that if I wish to static link that I need this other threadmon.dll as well. Has it been suggested that structured exceptions can perform the same function on win32 with a finally() handler from within thread_proxy? This is a gurantee of invocation. The cleanup attempt could though win32 doco states cleanup is guranteed, though couldnt a lib global could easily handle main thread cleanup? Regards Martin Johnson

2 1

Thread library: get thread id
by olivier_debels <olivier_debels＠yahoo.com> 17 Dec '02

17 Dec '02

I'm using boost_1_29_0 and need to get the id of the current thread. Is there a way to get the id of the current thread using boost? I don't see how I can access the current thread (Is there some sort of current function?). If I could access the current thread, I could write a getId() function (in a class subclassed from boost::thread). Only problem there is the return type. It seems the id is an int in windows environment, a pthread_t in POSIX and a MPTaskID in MACOS. Those last two seems to be pointers to struct. I guess it is possible to cast all of them to long (using reinterpret casting), but maybe I am overlooking something then. thanks, Olivier.

4 7