- Boost - lists.boost.org

Problem with multi_index_container
by Ilya Shvetsov 29 Oct '04

29 Oct '04

Hi All! This program demostrate the problem <code> #include <boost/multi_container_index.hpp> #include <boost/multi_index/ordered_index.hpp> int main(int argc, char* argv[] typedef boost::multi_index::multi_index_container<int> int_set; int_set s; int_set::iterator itr = s.begin(); ++itr; return 0; } </code> Program hangs up in increment if s is empty. I undestand that this is my problem. But may be "assert" can be inserted for such cases -- Ilya Shvetsov KranX Productions

2 2

C++ library with matrix inversion?
by Bronek Kozicki 28 Oct '04

28 Oct '04

I know that question is little off-topic here, but I hope there is someone who can answer it. Could you recommend some C++ library for matrix calculation that supports matrix inversion? Blizt++ does not have it, neither boost.ublas. Or I missed it. Ideas? TIA B.

8 8

RE: [boost] Circular log buffer
by Ferdinand Prantl 28 Oct '04

28 Oct '04

Hi Darren, > From: boost-bounces(a)lists.boost.org > [mailto:boost-bounces@lists.boost.org] On Behalf Of Darren Cook > > I'm finding an increasing need for an in-file memory log: > specifically when my program asserts I want to see where it has been. > > There are two problems with normal logging, i.e. appending to > a disk file: > * The files get very, very big (I've filled my home > partition before) > * All that writing to disk slows the application down noticeably. Usually you have to suffer the pain of the disk-writes because you need the log-file to analyze the reason of some crash and then the unflushed memory buffer to loose would not help much... The problem of disk space is normally solved by log rotating (parameters line maxlogfilesize and maxlogfilecount of a usual logging framework). If the logging causes unacceptable delays and is still required and cannot be moved to other parts of code with more details, I could imagine logging into a shared memory and a separate logging process flushing this memory to disk. I have not seen an ostream-like implementation for this, however. But circular buffer should be somewhere in the boost-sandbox. Ferda

2 1

BGL bidirectional adjacancy_list problem
by Hugo Duncan 28 Oct '04

28 Oct '04

We just updated to a recent cvs version of boost, and an algorithm that was taking three minutes is now running for over an hour. I think there are two problems, but am not at all sure of the analysis: The proximate cause seems to be the correction of the remove_edge function for bidirectional adjacancy_list based graphs in around June. Trying to see how to speed things up, I have run into these problems i) remove_edge is now O(E), due to the edge properties being stored on the graph rather than on a vertex. ii) removing properties from the edges changes nothing, a container of no_property's is still maintained. As I am not too familiar with the adjacancy_list implementation, can anyone confirm this? Hugo

2 5

Why no conversion from weak_ptr<T> to bool?
by Jonathan Biggar 28 Oct '04

28 Oct '04

It seems straighforward, it would just return !expired(). Any reason why not? -- Jonathan Biggar jon(a)levanta.com

2 1

Re: Boost Digest, Vol 894, Issue 1
by Michael Stevens 28 Oct '04

28 Oct '04

Hi Bronek, On Thursday 28 October 2004 08:42, Bronek Kozicki <brok(a)rubikon.pl> wrote: > I know that question is little off-topic here, but I hope there is > someone who can answer it. Could you recommend some C++ library for > matrix calculation that supports matrix inversion? Blizt++ does not have > it, neither boost.ublas. Or I missed it. Ideas? Linear algebra libraries generally avoid such inverse functions. This is because there are many algorithms possible depending on the properties of the matrix. The procedure is generally in two steps: decompose (factorise) the matrix in to simpler factors solve the linear equations using the factors Generally the above is done with LAPACK Have a look at "Linear Algebra with uBLAS" http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?Linear_Algeb… As mentioned there uBLAS also has an experimental "lu.hpp" which provide the LU factorisations (for square non-singular matrices) directly. These functions are experimental and without documentation so any feedback would be appreciated. Michael -- ___________________________________ Michael Stevens Systems Engineering Navigation Systems, Estimation and Bayesian Filtering http://bayesclasses.sf.net ___________________________________

1 0

[Q]: What is the recommended logging facility for boost: log4cpp or ??
by Francis ANDRE 28 Oct '04

28 Oct '04

Hi I am wondering what is the logging facility recommended by the boost community: We are using log4cpp and are quite satified for now of its functionnalities. But as usual, it is a separate package, not anymore maintained and out of the boost foundry... What 's your opinion on this subjet?? Cheers Francis -- A good friend will come bail you out of jail.......... but, a true friend....will be sitting next to you saying: "...that was fun."

15 43

[thread] doc formatting
by Vladimir Prus 28 Oct '04

28 Oct '04

There is a minor problem with Boost.Thread docs. On http://www.boost.org/regression-logs/cs-win32_metacomm/doc/html/libraries.h… the "Boost.Thread" header is in bold, while is different from names of all other libraries. On http://www.boost.org/regression-logs/cs-win32_metacomm/doc/html/threads.html "Boost.Thread" is also in bold. That is causes by the declaration of Boost.Thread entity in libs/doc/thead/entities.xml file. Is it possible to either change the definition of the entity of its usage? At least, the <title> should not be in bold. - Volodya

1 0

Boost libraries naming convention
by Tony Juricic 28 Oct '04

28 Oct '04

Just checking on my understanding of the motivation for current library names, prompted by some work on updating Visual Studio IDE projects. I assume that Boost version, build (i.e. debug or release) and CRT specific names like: libboost_serialization-vc71-mt-1_32.lib, libboost_serialization-vc71-mt-gd-1_32.lib and similar are intended for binary distribution of libraries, where end user is not even expected to use bjam, but simply link with the library that she needs. OTOH, the executables link with version, build and CRT-linkage independent library names like: libboost_serialization.lib. I assume that this is done simply for build convenience, since library is nothing else but a packaged set of object modules, whatever name you give to it. In fact, when I build boost with VC++ 7.1, using command: bjam -sTOOLS=vc-7_1<RETURN> (VC++ 7.1 environment variables are assumed set at this point) in my "...bin\boost\libs\test\build\libboost_test_exec_monitor.lib\vc-7_1\debug\th reading-multi" directory I see both libboost_test_exec_monitor-vc71-mt-gd-1_32.lib and libboost_test_exec_monitor.lib libraries, and, as I would expect, they are exactly identical. The *only* case where both libraries are *not* built is for "runtime-link-static\threading-multi" build, where I see only version-specific library, such as libboost_test_exec_monitor-vc71-mt-sgd-1_32.lib, for example. Am I missing some deep rationale behind the build logic and names given to the libraries, or? Tony

2 1

boost::signals - member-functions as slots
by Dirk Gregorius 27 Oct '04

27 Oct '04

Hi, I have a short question regarding the boost::signals library. Is it possible to connect a member function to a signal or must the passed slot be either static or be a function object? If yes what would be the syntax to pass the function and the associated object reference? -Dirk

3 2