April 2005 - Boost-users - lists.boost.org

RE: [Boost-users] RE: Boost thread problem
by Sliwa, Przemyslaw (London) 18 Apr '05

18 Apr '05

The CFXTrade is a dummy class. Here is the simple code, Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Ben Hutchings Sent: 18 April 2005 17:04 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] RE: Boost thread problem Sliwa, Przemyslaw (London) wrote: > Wow, wow, wow, > > Guys, this was my private program. I simply did not have access to the private email account from work. > As you have probably seen the issue was not with the CFXTrade class but rather with the vectro<T> one. I very much doubt that there's a bug in the std::vector implementation you're using. More likely CFXTrade doesn't meet the requirements for vector element types (CopyConstructible and Assignable). But and queue uses deque by default, and deque imposes the same requirements, so even if your program works on your current compiler it could break if you use any other compiler, or a different version, or even different options. Of course, we can't really tell if you don't post the missing code. Ben. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

2 1

RE: [Boost-users] Re: MultiIndexSet invariant test failure
by Manges, Ross G 18 Apr '05

18 Apr '05

Joaquin M Lopez Munoz wrote: > You've got to implement semantics equivalent to that of > the < relationship between numbers: > > http://www.sgi.com/tech/stl/StrictWeakOrdering.html Ah ha! After some serious unit testing, I uncovered a case in my sorting logic that does not comply with the strict weak ordering requirement. I have now fixed the logic, and I am no longer getting the invariant assert! Sorry to send you all on a "wild goose chase". I did consequently run into a SAFE_MODE related exception where an iterator to one of the indices was going invalid after an 'erase()' was called, and I'm still having problems with shared_ptrs, but I'll save the details for another discussion. Thanks for all of your help!

1 0

RE: [Boost-users] RE: Boost thread problem
by Sliwa, Przemyslaw (London) 18 Apr '05

18 Apr '05

I see the mailing program did not attach the code. Here it is: #pragma once #include <string> #include <fstream> class CFXTrade { public: CFXTrade(); CFXTrade(int, const std::string&, double); virtual ~CFXTrade(); public: int m_TradeID; std::string m_Name; double m_Amount; public: // Write the attributes to file void getWrittenToDisk(std::ofstream&); }; #include ".\fxtrade.h" #include <cmath> using namespace std; CFXTrade::CFXTrade() { } CFXTrade::CFXTrade(int nTradeID, const std::string& nName, double nAmount) : m_TradeID(nTradeID), m_Name(nName), m_Amount(nAmount) {} CFXTrade::~CFXTrade() {} // Write the attributes to file void CFXTrade::getWrittenToDisk(ofstream& nOutput) { nOutput << m_TradeID << ";" << m_Name << ";" << m_Amount << endl; } #include "FXTrade.h" #include <string> #include <sstream> #include <vector> #include <queue> #include <vector> #include <iostream> #include <cmath> #include <exception> #include <boost/random.hpp> #include <boost/thread/thread.hpp> #include <boost/thread/mutex.hpp> #include <boost/bind.hpp> using namespace std; using namespace boost; namespace write_namespace { mutex file_mutex; ofstream out; const size_t dimension(10000); const size_t size(10); const unsigned int number_of_threads(10); } queue<CFXTrade> TradeVector; namespace read_namespace { mutex in_file_mutex; ifstream in; } void write_to_disk(const vector<CFXTrade>& nVector) { using namespace write_namespace; mutex::scoped_lock lock(file_mutex); for(vector<CFXTrade>::const_iterator i_ptr=nVector.begin(); i_ptr!=nVector.end(); ++i_ptr) { out << i_ptr->m_TradeID << ";" << i_ptr->m_Name << ";" << i_ptr->m_Amount << endl; } } auto_ptr<CFXTrade> getTradeFromString(string& c_tmp) { int i_TradeID; string c_Name, c_sub; string::size_type pos; double d_Amount(0.0); pos = c_tmp.find(";"); c_sub = c_tmp.substr(0, pos); i_TradeID = atoi(c_sub.c_str()); c_sub = c_tmp.substr(++pos, 3); c_Name = c_sub; c_sub = c_tmp.substr(pos+4, 3); d_Amount = atof(c_sub.c_str()); d_Amount = 100.; return auto_ptr<CFXTrade>(new CFXTrade(i_TradeID, c_Name, d_Amount)); } void read_from_disk(queue<CFXTrade>& nTradeVector) { using namespace write_namespace; using namespace read_namespace; string c_tmp; try { mutex::scoped_lock lock(in_file_mutex); for(size_t i=0; i<dimension; ++i) { getline(in, c_tmp); TradeVector.push(*getTradeFromString(c_tmp) ); } } catch(exception& e) { mutex::scoped_lock lock(in_file_mutex); cout << "A standard exception caught from " << __LINE__ << endl; cout << e.what() << endl; } } bool generate_file() { using namespace write_namespace; vector< thread* > ThreadVector(number_of_threads); mt19937 mt2; random_number_generator<mt19937> std_rng(mt2); int TradeIDVector[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; char* NameVector[] = {"ABC", "BCD", "CDE", "DEF", "EFG", "FGH", "GHI", "HIJ", "IJK", "JKL"}; double AmountVector[] = {10., 20., 30., 40., 50., 60., 70., 80., 90., 11.}; vector< vector<CFXTrade> > TradeVector(number_of_threads); for(int i=0; i<number_of_threads; ++i) { for(int j=0; j<dimension; ++j) { TradeVector[i].push_back(CFXTrade(TradeIDVector[std_rng(size)], string(NameVector[std_rng(size)]), AmountVector[std_rng(size)])); } } out.open("file.txt"); for(int i=0; i<number_of_threads; ++i) { ThreadVector[i] = new thread(bind(&write_to_disk, TradeVector[i])); } for(int i=0; i<number_of_threads; ++i) { ThreadVector[i]->join(); } for(int i=0; i<number_of_threads; ++i) { delete ThreadVector[i]; } out.close(); return true; } bool read_trades() { using namespace write_namespace; using namespace read_namespace; vector<thread*> ReadThreadVector(number_of_threads); in.open("file.txt"); try { for(int i=0; i<number_of_threads; ++i) { ReadThreadVector[i] = new thread(bind(&read_from_disk, TradeVector)); } for(int i=0; i<number_of_threads; ++i) { ReadThreadVector[i]->join(); } for(int i=0; i<number_of_threads; ++i) { delete ReadThreadVector[i]; } } catch(exception& e) { cout << "A standard exception caught from " << __LINE__ << endl; cout << e.what() << endl; return false; } in.close(); return true; } void classified() { using namespace read_namespace; try { generate_file(); read_trades(); } catch(exception& e) { cout << e.what() << endl; exit(EXIT_FAILURE); } catch(...) { cout << "Unknown exception occured" << endl; } cout << TradeVector.size() << endl; } int main() { classified(); } -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Ben Hutchings Sent: 18 April 2005 17:04 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] RE: Boost thread problem Sliwa, Przemyslaw (London) wrote: > Wow, wow, wow, > > Guys, this was my private program. I simply did not have access to the private email account from work. > As you have probably seen the issue was not with the CFXTrade class but rather with the vectro<T> one. I very much doubt that there's a bug in the std::vector implementation you're using. More likely CFXTrade doesn't meet the requirements for vector element types (CopyConstructible and Assignable). But and queue uses deque by default, and deque imposes the same requirements, so even if your program works on your current compiler it could break if you use any other compiler, or a different version, or even different options. Of course, we can't really tell if you don't post the missing code. Ben. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

1 0

boost version 1.33
by Perepelitsa Roman 18 Apr '05

18 Apr '05

When will be boost version 1.33 released? Thanks in advance, Roma.

3 3

RE: [Boost-users] RE: Boost thread problem
by Sliwa, Przemyslaw (London) 18 Apr '05

18 Apr '05

The strange thing is -> works fine on a single processor Linux SUSE 9.1 but gives segmentation fault on a double processor machine with the same OS. I do not understand this. Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Ben Hutchings Sent: 18 April 2005 17:04 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] RE: Boost thread problem Sliwa, Przemyslaw (London) wrote: > Wow, wow, wow, > > Guys, this was my private program. I simply did not have access to the private email account from work. > As you have probably seen the issue was not with the CFXTrade class but rather with the vectro<T> one. I very much doubt that there's a bug in the std::vector implementation you're using. More likely CFXTrade doesn't meet the requirements for vector element types (CopyConstructible and Assignable). But and queue uses deque by default, and deque imposes the same requirements, so even if your program works on your current compiler it could break if you use any other compiler, or a different version, or even different options. Of course, we can't really tell if you don't post the missing code. Ben. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

1 0

A Simple Problem: File Transfer...
by Hossein Haeri 18 Apr '05

18 Apr '05

Dear all, If you were assigned to write a file transfer utitlity "in C++", what would your suggestions be? More specific, I've got two IP's (perhaps plus their ports), I've got to send one file from one to the other. Which parts of the Boost would you use, if any? TIA, --Hossein Send instant messages to your online friends http://uk.messenger.yahoo.com

2 1

RE: [Boost-users] RE: Boost thread problem
by Sliwa, Przemyslaw (London) 18 Apr '05

18 Apr '05

Wow, wow, wow, Guys, this was my private program. I simply did not have access to the private email account from work. As you have probably seen the issue was not with the CFXTrade class but rather with the vectro<T> one. Regards, Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of C. Michailidis Sent: 16 April 2005 09:12 To: boost-users(a)lists.boost.org Subject: [Boost-users] RE: Boost thread problem > I think the bug might lie in the copy-assignment operator for CFXTrade. > std::vector<CFXTrade> may use that whereas std::queue<CFXTrade> > probably will not. This is a very keen observation Ben! Whether or not this is the case, it certainly is motivation to make sure that sample code posted to the list is free from all references to proprietary code. In addition, I would say that before submitting code samples to boost-users we should make every effort possible to simplify the sample to the bare minimum required to reproduce the bug/problem in question. We should also consider that by submitting proprietary code to the list, we may be in violation of confidentiality agreements had w/ our employers or customers. Remember, this list is archived and could (theoretically) be the kind of proof an employer/customer needs to fire/sue you. After all, the boost-users list is about helping users w/ boost - not with 3rd party (and possibly proprietary) code. I think we all know that the quality of open-source peer-reviewed codes can be dramatically higher than that of closed-source; can boost-users really afford to help track down a multitude of problems in a potentially huge number of proprietary apps/libs? Oh, and always use protection when having promiscuous sex! Signed, a hard-liner ;-) _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

2 1

RE: [Boost-users] RE: Boost thread problem
by Sliwa, Przemyslaw (London) 18 Apr '05

18 Apr '05

Why not? AFAIK GSL, one of the most important open source numerical libraries uses double precision, so I do not see the point why you say double is not good enough for scientific applications. Regards, Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Mauricio Gomes Sent: 17 April 2005 04:12 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] RE: Boost thread problem Ben is right, you simply can't use double for any serious financial/scientific calculation. And indeed I hope someday an arbitrary math precision library be part of the standard C++. Look for a standard for the place where this is going to be used. If there isn't one. Use an arbitrary math precision library anyway. Take a look at mapm: http://www.cuj.com/documents/s=8020/cuj0111ring/ http://www.tc.umn.edu/~ringx004/mapm-main.html There are other comercial alternatives, a quick google search produced this: http://www.boic.com/b1mnum.pdf http://www.boic.com/numintro.htm Hope this helps, Mauricio Gomes Pensar Digital phone: 55-11-4121-6287 mobile: 55-11-8319-9610 http://pensardigital.com On Apr 15, 2005, at 2:37 PM, Ben Hutchings wrote: > Sliwa, Przemyslaw (London) wrote: >> Thanks, Ben, what should I use for representing the money amounts? A >> boolean :)) ? > > If you're dealing with other people's money then there are regulations > that specify the required intermediate precision and rounding rules, > which may vary from place to place. I believe you'll need to use > fixed-point decimal fractions, perhaps implemented as integer counts > of hundredths or ten-thousandths of a currency unit. I've not done > this myself so I don't know the details, but I'm fairly sure that > getting this wrong can put you on the wrong side of the law. > > Ben. > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

2 1

Re: You provide gun, I provide foot.
by C. Michailidis 18 Apr '05

18 Apr '05

> In any case, the boost libs still rock! Maybe I'll dig a little deeper and > make a proposal; after April 15th of course! After investigating various ways to solve my difficulties from previous posts (see http://lists.boost.org/boost-users/2005/04/11119.php and http://lists.boost.org/boost-users/2005/04/11120.php) I've settled on the following solution. I believe the changes contained in the following patch (to http://www.boost.org/libs/thread/example/recursive_mutex.cpp) enable more traditional/procedural mutex operation (i.e. un-scoped locks). I think my COM component is saved! diff -u recursive_mutex.cpp recursive_mutex_key.cpp --- recursive_mutex.cpp Mon Apr 18 08:52:07 2005 +++ recursive_mutex_key.cpp Mon Apr 18 09:03:03 2005 @@ -13,19 +13,27 @@ #include <boost/thread/thread.hpp> #include <iostream> +#define recursive_mutex_key boost::detail::thread::lock_ops<boost::recursive_mutex> + class counter { public: counter() : count(0) { } int add(int val) { - boost::recursive_mutex::scoped_lock scoped_lock(mutex); + int retval; + recursive_mutex_key::lock(mutex); count += val; - return count; + retval = count; + recursive_mutex_key::unlock(mutex); + return retval; } int increment() { - boost::recursive_mutex::scoped_lock scoped_lock(mutex); - return add(1); + int retval; + recursive_mutex_key::lock(mutex); + retval = add(1); + recursive_mutex_key::unlock(mutex); + return retval; } private: Feedback greatly appreciated, Dino P.S. Taxes all done, ya!

1 0

RE: [Boost-users] Lapack Bindings
by Lyon, Richard 18 Apr '05

18 Apr '05

Hi, A good alternative is to call the ATLAS CLAPACK/CBLAS DLL's directly from your c/c++ code. The only reason for using another layer on top of these libraries is too assist with the construction of the matrices, arithmetic operations, marshalling and I/O. You will need to look closely at the BLAS/LAPACK documentation at netlib to understand the various API's. With this approach you are not locked into c/c++. Calling CLAPACK/CBLAS from C# works well. ATLAS supports both row and column major type matrices. The matrix must be marshalled as a continuous array. Initially the CLAPACK/CBLAS interface looks complex, but after the initial hurdle is passed life is pretty straight forward. Looked and BLITZ and TNT but decided there seemed to be too much hype and not much documentation. LAPACK++ might be worth a try. Calling ATLAS CLAPACK/CBLAS DLL's directly works fine for me, but it really depends upon your requirements. Regards Richard Lyon -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Franz Mueller Sent: Sunday, 17 April 2005 9:15 AM To: boost-users(a)lists.boost.org Subject: [Boost-users] Lapack Bindings What about the future of the bindings, is there someone actively extending it? Is it actually worth using boost ublas/lapack-bindings or is there a better alternative? (TNT/JAMA, blitz, gmm,...) Thanks for your help, Franz _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0