- Boost-users - lists.boost.org

[bjam] using bjam as a build tool
by Ovanes Markarian 13 Feb '08

13 Feb '08

Hello bjam developers! I currently read through the docs for bjam and try to build my first project using. We currently need an automated build system in the current project to run directly from the version control system. The doc which I read currently leaves many items open which I do not understand, despite the fact I see bjam as a very cool thing! far superior as make;) These questions are open: 1. From the userman.pdf I learned that there is a possibility to use modules. Is there any location where I can what the modules do or how they work. Since such examples are not very well to understand: userman.pdf page 47-48 ------- "Accessing environment variables Many users would like to use environment variables in Jamfiles, for example, to control location of external libraries. In many cases you better declare those external libraries in the site-config.jam file, as documented in the recipes section. However, if the users already have the environment variables set up, it's not convenient to ask them to set up site-config.jam files as well, and using environment variables might be reasonable. In Boost.Build V2, each Jamfile is a separate namespace, and the variables defined in environment is imported into the global namespace. Therefore, to access environment variable from Jamfile, you'd need the following code: import modules ; local SOME_LIBRARY_PATH = [ modules.peek : SOME_LIBRARY_PATH ] ; exe a : a.cpp : <include>$(SOME_LIBRARY_PATH) ;" ------- 2. As I understood I can use user-config.jam to make default configurations for my project. I copied the user-config.jam into the project root, but it is not read, at least bjam issues warnings that toolset was not set and msvc is the default one. How can I specify in user-config.jam a default toolset? Or is the user-config-jam a file to declare available toolset, but toolspecification must be somewhere else? My user-config.jam contains: import toolset : using ; using msvc ; 3. How is it possible to specify the source location to 3d party lib (libpath)? I can't build a project with bjam v2 which uses boost libs. I don't want to enumerate each lib which my project depends on, but submit to the linker the lib path. My file content is: import modules ; local BOOST_STAGE_INCLUDE = [ modules.peek : BOOST_STAGE_INCLUDE ] ; local BOOST_STAGE_LIBS = [ modules.peek : BOOST_STAGE_LIBS ] ; exe MTSentry : main.cpp <source>$(BOOST_STAGE_LIBS) : <include>$(BOOST_STAGE_INCLUDE) <threading>multi ; The resulting error is: bjam.exe --v2 --toolset=msvc error: at Jamroot:7 error: properties found in the 'sources' parameter for ./MTSentry Removing the <source> option will build the app, but the linking fails, since boost/thread lib is not found. How can I specify a source directory for libs? With Kind Regards, Ovanes

2 3

OSI Approval
by KSpam 12 Feb '08

12 Feb '08

In perusing the list of OSI approved licenses today, I noticed that Boost is finally listed: http://www.opensource.org/licenses/alphabetical http://www.opensource.org/licenses/bsl1.0.html Although the approval process seems to be incredibly long and painful, it is great to see the Boost Software License officially recognized by OSI!

1 0

XML Serialization and the Generation of a W3C Schema
by Joseph Fradley 12 Feb '08

12 Feb '08

Hello, I've seen a few postings on how a XSD generating archiver would be written and a posting for a google code project on it. But I can't seem to find any implementations. Does anybody know of any implementations? Joe

2 1

[Regex] "Truly" incremental search?
by Eugene M. Kim 12 Feb '08

12 Feb '08

Hello, The current "incremental search" code needs to resume search where the a possible match starts, and for certain expressions the worst-case time (i.e. input bytes come in one at a time) complexity becomes O(n^2) where n is the number of characters that match the expression: Regex: omg.*wtf Pass 1 input: o Pass 1 input to matching: o Pass 1 verdict: Partial, resume at offset 0. Pass 2 input: m Pass 2 input to matching: om Pass 2 verdict: Partial, resume at offset 0. Pass 3 input: g Pass 3 input to matching: omg Pass 3 verdict: Partial, resume at offset 0. ... Pass n-1 input: t Pass n-1 input to matching: omg...wt (length n-1) Pass n-1 verdict: Partial, resume at offset 0. Pass n input: f Pass n input to matching: omg...wtf (length n) Pass n verdict: Match found and starts at offset 0. Total number of byte matches = n*(n+1)/2, i.e. O(n^2) I am looking for a way to feed the input bytes only once, in order to reduce the time complexity to approximately O(n). Of course I will lose the ability to recover submatches, but the ability is unnecessary for this particular application that I am considering. I cannot break the regex into two parts ("omg" and "wtf") either because it is specified as an input to the program. Eugene

5 6

bounded_matrix initialization
by Barberi 12 Feb '08

12 Feb '08

A boost newbie asks, how can I initialize a bounded_matrix? I tried various things like this: bounded_matrix<double, 2, 2> rot = list_of( 1.) (2.) (3.) (4.); and got the error message 'no overloaded function takes 2 arguments'. So far, I'm reduced to using copy assignment from a c array, but there's got to be a better way. What am I missing? BTW, I'm using MSVC++ 2008, with the MSVC++ 8 boost libraries.

1 0

[asio] Synchronization & infinite wait
by Sergei Politov 12 Feb '08

12 Feb '08

I have two questions about Boost.Asio library: 1) Is it safe to invoke deadline_timer::expires_from_now in thread different from thread where io_service is running? For instance: =========================================== void print(const boost::system::error_code& e) { std::cout << "Hello, world:" << e << "!\n"; } void execute(boost::asio::deadline_timer & timer) { timer.expires_from_now(boost::posix_time::seconds(0)); } int main() { boost::asio::io_service io; boost::asio::deadline_timer t(io, boost::posix_time::seconds(1)); t.async_wait(print); boost::thread th(boost::bind(&execute, ref(timer))); io.run(); return 0; } ============================================ 2) Are there any type of objects similar to events from WINAPI? I.e. these objects is not signalled (handler is not invoked) until they are signalled manually. Obviously such objects could be emulated using deadline_timer. But it would be great to have such objects itself.

1 0

Re: [Boost-users] Unresolved symbol in boost::filesystem lib with MSVC++ 2005 Pro
by Peter Jansson 12 Feb '08

12 Feb '08

Hi, I have the same issue as described in this thread and I am very much interested to hear if anyone have any suggestion on how to solve it. Best regards, Peter Jansson Sweden

1 0

Static build on darwin
by Thomas Hauk 12 Feb '08

12 Feb '08

Hi all, I'm having trouble with using Boost.Thread (1.34.1) on Mac OS X (10.4.11/ppc/Xcode 2.3). I have an Xcode project ("BSD command line tool") that links to what I believe is the static build of Boost.Thread (libboost_thread-mt- d-1_34_1.a). However, when Xcode goes to run the executable, it gives the error: > Running… > dyld: Library not loaded: bin.v2/libs/thread/build/darwin/debug/ > threading-multi/libboost_thread-mt-d-1_34_1.dylib > Referenced from: /path/to/my/executable/executablename > Reason: image not found > Program received signal: "SIGTRAP". If I run the executable inside an X11 shell, I get a similar error. If I add the path to the .dylib to the $DYLD_LIBRARY_PATH environment variable, I can run the executable. Which makes me wonder if somehow this .a file is really a .dylib. What's also strange is that although the build directories for all the other libraries have the same structure (e.g. filesystem/build/ darwin/debug/link-static/libboost_filesystem-d-1_34_1.a), the thread directory has a threading-multi directory under the link-static directory (meaning the file I am trying to link to is in thread/build/ darwin/debug/link-static/threading-multi/libboost_thread-mt-d-1_34_1.a). I'm building all the libraries with bjam 3.1.15 and the command line "bjam --toolset=darwin". FWIW, I'm also using the Boost filesystem library with this executable, without any compile, link or run-time errors. Any help/ideas? Also: is there any plan to distribute the Boost libraries for Mac OS X in a similar way to how they are distributed for Windows (i.e. installer that has pre-built library binaries)? T -- Some people, when confronted with a problem, think "I know, I'll use regular expressions." Now they have two problems. --Jamie Zawinski, in comp.lang.emacs

1 0

[serialization] exceptions are not caught
by Krishanu Debnath 12 Feb '08

12 Feb '08

Hello, This is a platform specific problem, platform is RHEL4. We use boost::serialization code, and linked with their static library. And the problem we are facing that all of our exception handler codes are not working any more i.e. catch handler is not called if an exception is thrown. We are not even calling the boost::serializer code. If we remove serializer code everything works fine. Serializer code works pretty fine in suse platform. Now my question is how come exception handler codes are not executing just because we are linking with serializer library. Any insight will be helpful. Thanks, Krishanu

1 0

[filesystem] bug in wpath with utf8 locales
by Filip Konvička 12 Feb '08

12 Feb '08

Hi, there is a problem with wpath when the system locale is UTF-8. My setup is g++ 4.2.3 --enable-clocale=gnu on a 2.6.18-5-686 GNU/Linux, boost 1.34.1, locale "cs_CZ.UTF-8". The problem seems to be in translating wstring into native strings inside Filesystem. The error I'm getting (and others too, see for example "Problem with exists() and wpath" and "Bug in filesystem library?" on this list) is boost::filesystem::wpath::to_external conversion error I've seen some discussions on the boost-dev list dealing with this issue, and to me this looks like a serious bug in wpath implementation. Until this is resolved, I think that we need to #ifdef this and use char-based paths on Linux. Are there some plans to fix this? I'm willing to help if that is needed. Thanks, Filip

1 0