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

Macintosh Carbon Thread Error with Boost 1.30.0
by Bobby Thomale 14 Apr '03

14 Apr '03

Hey guys, I just tried to get the new Boost 1.30.0 release to work on MacOS, using TARGET_CARBON (with threads turned on). I have a problem to report: there's a new "lightweight mutex" being used by the smart pointer library. There does not seem to be a lightweight mutex implementation for the "Carbon MP tasks" thread implementation. So - when I compile my code under Boost 1.30 I end up at: # error Unrecognized threading platform in lightweight_mutex.hpp when it tries to compile any of the smart pointer stuff. Help! :-) -- Bobby --------------------------------------------------------------------- Bobby Thomale Senior Software Developer Inoveon Corporation http://www.inoveon.com/ ---------------------------------------------------------------------

2 3

uBlas Lapack binds from Sandbox - pb with templ. specialisation
by geert_ceuppens 14 Apr '03

14 Apr '03

Toon & Kresimir, I've downloaded the lapack and blas binds from the boost sandbox, and I found out the hard way (again :^( ...) the 'features' of the VC6 compiler, i.e. the lack of support for template specialisation. The offending code is: (line nbr may differ due to versioning) *) boost\numeric\bindings\traits\algorithm.hpp(45): template < typename T > struct is_equal vs: template < typename T > struct is_equal< boost::numeric::ublas::matrix< T > > --> template class has already been defined as a non-template class *) libs\numeric\bindings\blas\test\blas.hpp(51): template < typename T > struct assign_multiplier vs: template < typename T > struct assign_multiplier< std::complex< T > > *) boost\numeric\bindings\traits\traits.hpp(46): template <typename V, typename T = typename V::value_type> struct default_vector_traits vs: template <typename V, typename T> struct default_vector_traits<V const, T> *) boost\numeric\bindings\traits\traits.hpp(112 & 114): This one gave more errors: template <typename MatrixT> inline typename matrix_traits<MatrixT>::pointer matrix_storage (MatrixT& m) { return matrix_traits<MatrixT>::storage (m); } error C2039: 'pointer' : is not a member of 'matrix_traits<M>' error C2146: syntax error : missing ';' before identifier 'matrix_storage' error C2433: 'pointer' : 'inline' not permitted on data declarations error C2501: 'pointer' : missing storage-class or type specifiers error C2059: syntax error : ';' error C2065: 'MatrixT' : undeclared identifier error C2065: 'm' : undeclared identifier 'matrix_storage' : missing storage-class or type specifiers syntax error : missing ';' before '<end Parse>' Any chance of providing non-specialised versions of these so I can use the VC6 compiler? Thanks - Geert

1 0

RE: [Boost-Users] command line parsing
by Jim.Hyslop 14 Apr '03

14 Apr '03

Joe Mariadassou [mailto:joemar@vcisnet.net] wrote: > The URL appears to be invalid. > joe > > -----Original Message----- > From: HartmutKaiser(a)t-online.de [mailto:HartmutKaiser@t-online.de] > Sent: Monday, April 14, 2003 3:04 AM > To: Boost-Users(a)yahoogroups.com > Subject: RE: [Boost-Users] command line parsing > > > As far as I remember Vladimir Prus has requested Boost review for his > program_options library. You can find it here: > > http://zigzag.cs.msu.su:7813/program_options > Works OK for me - maybe the problem is temporary. -- Jim

1 0

test
by iad929 13 Apr '03

13 Apr '03

plz ignore. Mohammed

1 0

RE: [Boost-Users] Unary predicate for shared_ptr == 'real' pointe r?
by dick.bridges＠tais.com 11 Apr '03

11 Apr '03

Thanks! In a sense, I can sympathize with your co-workers. I spent hours trying to get that predicate right - I kept screwing up the 'inner' bind. I keep hoping the more I work with the boost libraries the easier it will get. In my case, that's yet to be seen. ;) Thanks again. Tom Matelich <tmatelich@zetec To: "'Boost-Users(a)yahoogroups.com'" <Boost-Users(a)yahoogroups.com> .com> cc: Subject: RE: [Boost-Users] Unary predicate for shared_ptr == 'real' pointe r? 04/11/2003 01:55 PM Please respond to Boost-Users > From: Mark Storer [mailto:mstorer@cardiff.com] > Sent: Friday, April 11, 2003 1:51 PM > > The whole "new int" thing seems a little odd, but that's an > artifact of creating example code. precisely > What do your co-workers say? I feel like I'm missing something here. While the elegance of the double bind is appreciated, I've heard complaints about it being confusing. ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. ----------------------------------------------------------------------- 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] Unary predicate for shared_ptr == 'real' pointe r?
by Tom Matelich 11 Apr '03

11 Apr '03

> From: Mark Storer [mailto:mstorer@cardiff.com] > Sent: Friday, April 11, 2003 1:51 PM > > The whole "new int" thing seems a little odd, but that's an > artifact of creating example code. precisely > What do your co-workers say? I feel like I'm missing something here. While the elegance of the double bind is appreciated, I've heard complaints about it being confusing. ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. -----------------------------------------------------------------------

1 0

RE: [Boost-Users] Unary predicate for shared_ptr == 'real' pointe r?
by Mark Storer 11 Apr '03

11 Apr '03

The whole "new int" thing seems a little odd, but that's an artifact of creating example code. What do your co-workers say? I feel like I'm missing something here. --Mark Storer Software Engineer Cardiff Software #include <disclaimer> typedef std::disclaimer<Cardiff> Discard;

1 0

RE: [Boost-Users] Unary predicate for shared_ptr == 'real' pointe r?
by Tom Matelich 11 Apr '03

11 Apr '03

> From: dick.bridges(a)tais.com [mailto:dick.bridges@tais.com] > Sent: Friday, April 11, 2003 11:54 AM > > This is probably more of a C++ than an boost question but, since it > involves shared_ptr, I thought I'd try it here first. I've > got a vector of > shared_ptrS (e.g., vector<shared_ptr<T> > v;) and I need to > find the entry > given the 'real' pointer to the underlying T. Looping thru > the vector like > this: > > <snippet> > for( vector<shared_ptr<T>::iterator itr = v.begin(); itr != > v.end(); itr++) > { > if( itr->get() == real_this){...} > } > </snippet> > > works but seems klunky. I should be able to use [find_if( > v.begin(), v.end > (), ???)] but I can't work out the predicate. Any help or > "pointers" ;) > will be greatly appreciated. > Here's how I like to do it, though my co-workers sometimes complain: #include "boost/shared_ptr.hpp" #include "boost/bind.hpp" #include <vector> #include <iostream> int main() { typedef std::vector<boost::shared_ptr<int> > V; V vec; int* foo = new int; vec.push_back(boost::shared_ptr<int>(foo)); V::iterator iter = std::find_if(vec.begin(), vec.end(), boost::bind(std::equal_to<int*>(), boost::bind(&boost::shared_ptr<int>::get, _1), foo)); if(iter != vec.end()) std::cout << "Life is good" << std::endl; //don't delete foo!!! return 0; } Tom ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. -----------------------------------------------------------------------

1 0

Unary predicate for shared_ptr == 'real' pointer?
by dick.bridges＠tais.com 11 Apr '03

11 Apr '03

This is probably more of a C++ than an boost question but, since it involves shared_ptr, I thought I'd try it here first. I've got a vector of shared_ptrS (e.g., vector<shared_ptr<T> > v;) and I need to find the entry given the 'real' pointer to the underlying T. Looping thru the vector like this: <snippet> for( vector<shared_ptr<T>::iterator itr = v.begin(); itr != v.end(); itr++) { if( itr->get() == real_this){...} } </snippet> works but seems klunky. I should be able to use [find_if( v.begin(), v.end (), ???)] but I can't work out the predicate. Any help or "pointers" ;) will be greatly appreciated. TIA

1 0

RE: [Boost-Users] Boost random's min < max assert
by Glen Knowles 11 Apr '03

11 Apr '03

From: Jens Maurer >> Uniform int contains an assert if min < max. If I ask for a random between 1 >> and 1 I'm expecting a 1, not an assert. Is there some reason this was deemed >> inappropriate? > >When I wrote that code, I never thought somebody would want to generate >a constant 1 as a random number. I agree that this looks non-orthogonal >on first sight. I'll look into it. In my specific case I have a number of objects determined at run time and am randomly picking a number between one and the number of objects in order to select one. The workaround is to simply test for one before the call, but the assert was unexpected. Thanks, Glen [Non-text portions of this message have been removed]

1 0

RE: [Boost-Users] Problem with boost::pool_alloc
by scleary＠jerviswebb.com 11 Apr '03

11 Apr '03

> -----Original Message----- > From: Ilya Chvetsov(work) [mailto:chvetsov@kdlab.com] > > Hi! > I try to use boost::pool_alloc with STLport-4.5-0119 snapshot. > STLPort call pool_alloc::allocate(const size_type n) with n = 0. > STLPort naitive allocator returns in such case NULL, but pool_alloc > tries return memory. Hmmm... You seem to have uncovered a corner case, not covered by the Standard requirements. I will change the pool_allocator (and fast_pool_allocator) to be able to handle this by returning NULL. > Method simple_segregated_storage<SizeType>::try_malloc_n( > void * & start, size_type n, const size_type partition_size) has > problem too if parameter n is 0. Problem is in while condition: > while (--n != 0)... This is correct; (n > 0) is a precondition for that function. -Steve

1 0

RE: [Boost-Users] Problems with iterator_adaptors
by Ben Hutchings 11 Apr '03

11 Apr '03

Olaf Petzold wrote: > I try to write a cyclic Iterator using boost::iterator_adaptors. > Unfortunally, the advance function doesn't work for negative distances > correct: > template<class Iterator> > class cycle_iterator_policies > { > public: > cycle_iterator_policies(const Iterator& begin, const Iterator& end) > : m_begin( begin ), m_end( end ) > { > assert( m_begin != m_end ); > } > .... > template<class IteratorAdaptor, class DifferenceType> > void advance( IteratorAdaptor& x, DifferenceType n ) { > n %= std::distance( m_begin, m_end ); <snip> The problem is that the result of % and / (and by implication %= and /=) is implementation defined if either operand is negative. Most implementations round the quotient towards zero and produce a remainder with the same sign as the quotient. Evidently you want to have a non-negative remainder always. The simplest solution, I think, is: DifferenceType length = std::distance( m_begin, m_end ); n %= length; if (n < 0) n += length;

1 0

Re: thread lib using workshop CC on solaris
by Christopher Currie 11 Apr '03

11 Apr '03

I tried to send this earlier using the newsgroup, but it seems to have been dropped on the floor. Sorry for the duplicate if it shows up. The latest version of boost has some support for some versions of this compiler. The appropriate jamfile is "sunpro-tools.jam" (Sunpro is an old name for this compiler.) Thus: cd $BOOST_ROOT/libs/thread/build bjam -sTOOLS=sunpro That said, your mileage may vary on this compiler. I have found that it is notoriously buggy, especially wrt templates, and boost code will often cause it heartache in obscure ways. My results have found: * Up-to-date patches for the compiler and the operating system are a must. Find them here: http://access1.sun.com/sundev/index.html * Both versions of this compiler that I have tested have issues with Boost.TypeTraits, generating loads of spurious "Anacronism" warnings wrt the typename keyword being absent or present. Unfortunately, sun has not provided a flag to silence these warnings, and adding typename to the locations it complains about breaks other compilers, notably g++. * Both versions of this compiler will have problems linking some Boost unit tests, due to odd template RTTI issues. Adding -sBUILD="<cxxflags>-instances=static" will sometimes fix this problem. I don't recommend making this flag a default, because of the code bloat that entails. * Sun CC 5.4 ( Workshop/Forte Developer/SunOne Studio 7 ) will successfully build Boost.Threads. * Sun CC 5.3 update 2 ( W/FD/SOS 6u2 ) will only build Boost.Threads in release mode. This compiler generates assertions when attempting to compile the debug libraries. * I have not tested any versions earlier than Workshop 6 update 2. "jenny_qzhang" <jenny_qzhang(a)yahoo.com> wrote in message news:b722dr+7f4@eGroups.com... > Hi, All > > I am working on porting an application from MSVC to solaris > workshop(the applicatoin is using boost thread lib). I could not > locate the makefile or jam file for this compiler. Is there anybody > did the same thing before? > > If you finially made it, can you share your experience, makefile > with me? Actually I tried to write a makefile myself, but it reported > a lot of errors related to template parameters :(. > > > Thanks very much! > > Qiu

1 0

Problems with iterator_adaptors
by Olaf Petzold 11 Apr '03

11 Apr '03

Hello, I try to write a cyclic Iterator using boost::iterator_adaptors. Unfortunally, the advance function doesn't work for negative distances correct: template<class Iterator> class cycle_iterator_policies { public: cycle_iterator_policies(const Iterator& begin, const Iterator& end) : m_begin( begin ), m_end( end ) { assert( m_begin != m_end ); } .... template<class IteratorAdaptor, class DifferenceType> void advance( IteratorAdaptor& x, DifferenceType n ) { n %= std::distance( m_begin, m_end ); DifferenceType rest = std::distance( x.base(), m_end ); if( n < rest ) std::advance( x.base(), n ); else std::advance( x.base() = m_begin, n - rest ); } .... private: Iterator m_begin; Iterator m_end; }; It seems to be an algorithmic problem for me. How can I corrected it? Thanks Olaf

1 0

Signals and noncopyable
by svanechel 11 Apr '03

11 Apr '03

Hi, Is there a reason why siganls aren't copyable? As far as I can tell signal base derives from noncopyable, but it references its implementation using shared_ptr. So I'm guessing in theory a signal is perfectly copyable. I must admit I came to this conclusion by briefly looking at the code, so I might be completely wrong. I want to store signals into a vector of function<void()>'s, together with bound arguments, to capture events and replay them later. Something similar to (simplified): class recorder { public: // this is the handler for some event // it is connected to another signal void some_event( int i) { events_.push_back( bind( ev, i ) ); } void replay() { vector< function<void()> >::const_iterator itr = events_.begin(); while ( itr != events_.end() ) { (*itr)(); ++itr; } } signal<void(int)> ev; private: vector< function<void()> > events_; }; recorder rec; // connect the recorder to the signals is uses ... // now replay the events rec.ev.connect( some_handler ); rec.replay(); The example above may no be entirely correct since I'm coding this directly without running it through a compiler, but I hope the intent is clear. If I remove the derivation from noncopyable from signal_base, a sigal can be assigned to a function, although I haven't tested it yet to see if it actually works (currently I do not have something I can start testing with). Is it safe to remove the noncopyable base class? If so can it be done in the next release of the boost distribution as well? Thanks in advance, Sven Van Echelpoel

1 0

Problem with boost::pool_alloc
by Ilya Chvetsov(work) 11 Apr '03

11 Apr '03

Hi! I try to use boost::pool_alloc with STLport-4.5-0119 snapshot. STLPort call pool_alloc::allocate(const size_type n) with n = 0. STLPort naitive allocator returns in such case NULL, but pool_alloc tries return memory. Method simple_segregated_storage<SizeType>::try_malloc_n( void * & start, size_type n, const size_type partition_size) has problem too if parameter n is 0. Problem is in while condition: while (--n != 0)... This simple program demonstrates a problem: #include <boost/pool/poolfwd.hpp> #include <boost/pool/pool_alloc.hpp> typedef boost::pool_allocator<int> alloc_t; int main(int argc, char* argv[]) { alloc_t alloc; alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); alloc.allocate(0); return 0; } How can I solve this problem? Thanks -Ilya

1 0

Boost random's min < max assert
by Glen Knowles 10 Apr '03

10 Apr '03

Uniform int contains an assert if min < max. If I ask for a random between 1 and 1 I'm expecting a 1, not an assert. Is there some reason this was deemed inappropriate? Thanks, Glen [Non-text portions of this message have been removed]

3 2

preprocessor: convert to capitals
by Toon Knapen 10 Apr '03

10 Apr '03

In the ublas forum we're having a discussion on the best way to provide signatures in C for fortran libraries (like BLAS and LAPACK). For linking C with fortran, you sometimes need to add an underscore to the function in C, but on other platforms fortran will store the function-name in capitals. Thus on different platforms, we should preprocess our source to take the platforms' convention into account and we thus would need a macro to convert an identifier automatically to uppercase. Is that possible ? toon -- Toon Knapen http://www.fft.be Free Field Technologies tel:+32 10 45 12 26 16 Place de l'Universite fax:+32 10 45 46 26 1348 Louvain-la-Neuve http://www.femtown.org Belgium

3 5

thread lib using workshop CC on solaris
by jenny_qzhang 10 Apr '03

10 Apr '03

Hi, All I am working on porting an application from MSVC to solaris workshop(the applicatoin is using boost thread lib). I could not locate the makefile or jam file for this compiler. Is there anybody did the same thing before? If you finially made it, can you share your experience, makefile with me? Actually I tried to write a makefile myself, but it reported a lot of errors related to template parameters :(. Thanks very much! Qiu

2 2

ublas matrix slice bug?
by Stephan Puchegger 08 Apr '03

08 Apr '03

Dear ublas developers! (I am using boost 1.30 & gcc 3.2) the following program: #include <boost/numeric/ublas/matrix.hpp> #include <boost/numeric/ublas/io.hpp> int main () { using namespace boost::numeric::ublas; matrix<double> m (4, 4); for (unsigned int i = 0; i < m.size1 (); ++ i) for (unsigned int j = 0; j < m.size2 (); ++ j) m(i, j) = 1; std::cout << m << std::endl; // multiply a slice of the matrix with 2.... project (m, slice (2, 1, m.size1()-2), slice (1, 0, m.size1()-2)) *= 2; // interesting result ... the slice has been multiplied with 4 std::cout << m << std::endl; } gives the following output: [4,4]((1,1,1,1),(1,1,1,1),(1,1,1,1),(1,1,1,1)) [4,4]((1,1,1,1),(1,1,1,1),(1,4,1,1),(1,4,1,1)) Why is the slice multiplied with 4 and not with 2?? Do I make some fundamental mistake? Everything works fine if I am using a matrix_vector_slice. My second (stupid?) question is the following: Is it intentional, that it is not possible to compile: #include <boost/numeric/ublas/matrix.hpp> #include <boost/numeric/ublas/io.hpp> int main () { using namespace boost::numeric::ublas; matrix<double> m (4, 4); for (unsigned int i = 0; i < m.size1 (); ++ i) for (unsigned int j = 0; j < m.size2 (); ++ j) m(i, j) = 4*i+j; std::cout << m << std::endl; swap(row(m,0), row(m,2)); std::cout << m << std::endl; } whereas matrix_row<matrix<double> > l(m, 0); matrix_row<matrix<double> > r(m, 2); swap(l, r); /* swap the rows */ works fine. MTL has support for a similar syntax: swap(rows(m)[0], rows(m)[2]); best regards and thanks for your help! Stephan

3 7

Spirit PDF or something
by Edward Diener 08 Apr '03

08 Apr '03

I would like to print out the Spirit documentation as a whole for offline study. However Boost 1.30 Spirit documentation comes as a series of HTML pages. Is there an easy way to print out the documentation rather than cycling through all of it manually and printing out a page at a time ?

2 3

condition::wait() - Alertable in Win32
by Simon Bailey 08 Apr '03

08 Apr '03

What is the latest thinking on making condition::wait() and condition::timed_wait() "alertable" in the Win32 environment? Please consider this a "me too" request for such a feature! I'm trying to implement a cross-platform (Unix, Windows) implementation of a Proactor pattern that manages asynchronous sockets. It seems that the way to implement this pattern with as much code common to both platforms as possible is to use 1. Overlapped sockets that call back (not post messages) in Windows. 2. Signals in Unix (similar to the Windows call backs but with less- clear semantics in my mind). For this scheme to work, the wait functions need to be alertable in Windows, but the Unix system shouldn't need any changes. >From the research that I have done so far, in my particular case, I would only need wait and timed_wait to be alertable, and it would be appropriate that they were always altertable - I would not need a way to enable or disable this feature.

1 0

Building (runtime-link-static) thread lib on vc7
by Taras Tielkes 07 Apr '03

07 Apr '03

Hi, Some newbie boost thread lib questions: 1) I used to build the 1.29 thread library with the runtime-link-static\threading-multi build configuration. This would yield (in release mode) a 467 KB import library in the destination directory. When I build boost 1.30 (with runtime-link-static\threading-multi), the target directory for the thread library contains: -a boost_thread.dll of 116 KB -boost_thread.lib import libarary of 27KB Programs compile against the import library have a runtime dependency on the boost thread dll (as I would expect from the size of the import .lib) What's the right way to compile a static-link, multi-threaded boost.threads library (using the vc7 toolset) in 1.30? 2) What happended to the threadmon.dll? Is it gone in 1.30? Thanks for any pointers..

1 0

RE: [Boost-Users] Re: Please, please help!!!
by scleary＠jerviswebb.com 07 Apr '03

07 Apr '03

> From: agiatojon [mailto:JonAgiato@nyc.rr.com] > > Thanks a lot for the info. I saw that but did not know exceptions > were disabled in some compilers by default, which as you suggested, I > found out this one does. I enabled them, and now it compiles. > > I have another question for you if I may. I attempted to simply > stream the numbers to the standard output but it complains that there > is no operator defined for this object. Is there a file I can include > that contains the overloaded output operator functionality? I am sure > this must be available, I am just not sure of its location. I know > that ublas uses the file "IO" in order to bring those definitions in. Make sure you're doing: std::cout << dev(); and not: std::cout << dev; If that isn't the problem, could you post a code sample? -Steve

2 1

specializing new lexical_cast
by Tom Matelich 07 Apr '03

07 Apr '03

I'm going to embark on it, but I thought I'd ask in parallel. One of the things about Terje's proposed lexical_cast was a lot of documentation of how to specialize it. That doesn't seem to be addressed in 1.30.0's version. I see in the documentation "Where non-stream-based conversions are required, lexical_cast is the wrong tool for the job and is not special-cased for such scenarios." I understand this sentiment, and I tried to search for discussions of this on the mailing list, but I did not conquer ASPN's search tool. The reason I'm asking is that we did some profiling and strtod kicked lexical_cast's butt for string -> double. I currently use lexical_cast in some template'd xml code, so I could overload my function, but I'd have to do it in about 5 spots due to an evolving interface. I'd like to just do it once for lexical_cast. I suppose I could use Terje's implementation, but I've stuck with boost canon for a long time now intentionally. Any suggestions? ----------------------------------------------------------------------- DISCLAIMER: Information contained in this message and/or attachment(s) may contain confidential information of Zetec, Inc. If you have received this transmission in error, please notify the sender by return email. -----------------------------------------------------------------------

4 4