January 2006 - Boost-users - lists.preview.boost.org

Re: [Boost-users] [test] How do you create test cases in a DLL
by Stewart Tootill 13 Jan '06

13 Jan '06

This may not help, and may not be the right way to go about it, but typically I get around this by putting all of the internal bits into a static lib, then linking that into the test binary. This also means that when I link the DLL, the test bits are left out, so I don't have to ship alot of redundant test code which the unused testInternals export would cause to be included in the final DLL. Admitedly, my DLLs are designed to be built in this way.. -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org]On Behalf Of Mark Snelling Sent: 12 January 2006 17:53 To: boost-users(a)lists.boost.org Subject: [Boost-users] [test] How do you create test cases in a DLL I have some internal (non-exported) classes in a DLL that I want to test. I'm trying to do this by exporting a testInternals() function that my tester application will add to it's test suite. The problem I'm having is that to use any of the BOOST_CHECK_??? etc macros I have to include the <boost/test/test_tools.hpp> file. This causes undefined symbol linker errors. Is there a way to do what I want to do? ________________________________________________________________________ This e-mail has been scanned for viruses by MessageLabs. ________________________________________________________________________ This e-mail has been scanned for viruses by MessageLabs.

1 0

Re: [Boost-users] [serialization] [wave] [spirit] Can't compile o ost cvs
by James Hughes 13 Jan '06

13 Jan '06

> > With reference to this fix, where can I get it or should I > > revert to 1.33.0 rather than 1.33.1 which I currently have? > > The patch Joel was talking about never got applied to the > 1.33.1 branch. > Seems you are referring to another problem, which wasn't > investigated yet. > So give us some time to figure out what went wrong, please. > > Regards Hartmut > > Sorry, they seemed very similar problems from the description. I'll hang loose for the moment then! I will go back to 33.0 to see if that compiles cleanly for arm (and post any significent resutls), and in the meantime keep an eye on things here for any news. Thanks. James

2 1

Assertion failure usign boost::format with Visual C++ 2005
by Jean-Sébastien Otis 12 Jan '06

12 Jan '06

Hello, Here is a sample that generates an assertion failure with Visual C++ 2005 using boost 1.33.1. A NULL ptr is passed to std::string::append() by boost::io::detail::mk_str(). //////////////////////////////////////////////////////////////////// #include <boost/format.hpp> int main(int argc, char* argv[]) { std::string str(""); // Empty string. boost::format frmt( "%s" ); std::string res = ( frmt % str ).str(); return 0; } //////////////////////////////////////////////////////////////////// This is due to all the assert statements added in the new "secure" STL that is shipped with VC2005. But still, I don't think passing NULL to string::append() is correct anyway... Regards. J.S.Otis Software Developper InnovMetric Software Inc.

1 0

[Boost.Build v2] vc 8.0 manifest problem
by Jaroslav Gresula 12 Jan '06

12 Jan '06

Running bjam.exe on the Jamroot listed below gives bin\msvc-8.0\debug\runtime-link-static\shared.dll.manifest : general error c1010070: Failed to load and parse the manifest. The system cannot find the file specified. <Jamroot> using msvc : 8.0 ; lib shared : file.cxx : <link>shared <runtime-link>static ; </Jamroot> I've noticed that removing <runtime-link>static or replacing it with <runtime-link>shared 'fixes' the problem. I use Boost.Build from the Boost CVS and MSVC 8.0 Professional Edition. -- Jarda

3 3

boost::gregorian::months : access value / serialize
by Oliver.Kowalke＠infineon.com 12 Jan '06

12 Jan '06

Hallo, I've to serialize the value of months into a databse. This requires to access the value (amount of months) stored inside of boost::gregorian::months. How can I access the internal value (nothing found in the docu)? If this is not possible which boost::date_time type should I in order to store a datetime duration consisting of the pair years and months (Oracle database type INTERVAL YEAR TO MONTH)? Regards, Oliver

2 1

Re: [Boost-users] [serialization] [wave] [spirit] Can't compile b oost cvs
by James Hughes 12 Jan '06

12 Jan '06

> -----Original Message----- > From: Peter Whaite [mailto:boost@whaite.ca] > Sent: 11 January 2006 17:49 > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] [serialization] [wave] [spirit] > Can't compile > boost cvs > > > Joel de Guzman <joel(a)boost-consulting.com> wrote: > > > > Ok, confirmed. I'm the culprit :( Someone submitted a patch > > for the is* functions to work on other POD types. Hartmut is > > reverting it now while I try to find a fix. > > OK thanks it compiles now. > > I'm still surprised this wasn't reported earlier as > serialization seems > to be one of the more popular libs. > With reference to this fix, where can I get it or should I revert to 1.33.0 rather than 1.33.1 which I currently have? Rgds James

2 1

Re: [Boost-users] Cross compiling boost libraries with gcc and using boost jam to crosscompile third-party library (Jambase problem)
by Stering Wolfram 12 Jan '06

12 Jan '06

Vladimir, >recently I was put into task to cross compile boost for ARM platform >with GCC. > >My building platform is i386 Linux with gcc and cross compiling to arm >Linux. > > I had the very same problem for my Coldfire cross-build (target "m68k-elf"), working under Windows XP using Cygwin. I didn't find a "standard way" of doing a cross-build of boost in its build documentation. >Standard way for this kind of compilation is, that name of cross >compiler (running under i386 but producing arm code) and all cross >binary utilities are prefixed by target plarform, so like arm-linux-gcc, >arm-linux-g++, arm-linux-ar, arm-linux-objcopy for arm-linux. > > This applies when using the GNU toolchain. But nowadays, most cross compilers are gcc based. >And this is causing a bit of problem with boost. > >I can force GCC and GXX, but still have a problem with ar and objcopy. > >I've tried to setup GCC_BIN_DIRECTORY but it didn't work. > > I did basically the same thing to tools/build/v1/gcc-tools.jam as you did, at least with respect to ar and objcopy being available as variables that can be set on jam command line. I then used "-sGXX=m68k-elf-g++ -sGCC=m68k-elf-gcc -sAR=m68k-elf-ar" on my build command line. Additionally, I added support for selecting the CPU model (architecture/instruction-set) by adding the following: (I don't know how/why this works, just imitated the respective block for other architectures). Perhaps, boost build system should support (kind of) multilib builds for cross compilation, in a similar way that gcc/g++ does when cross-compiling it for targets where this applies? This enabled me to add "-sBUILD=debug release <runtime-link>static <optimization>size <architecture>m68k <instruction-set>5200" to the build command line for a Coldfire cross-build. It even somehow worked, as gcc got called with the correct -m option... # Coldfire and other m68k (WS, Tue Dec 20 16:46:10 2005). .ARCH = <architecture>m68k <architecture>native ; flags gcc CFLAGS <architecture>m68k/<instruction-set>default : -m5200 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>68000 : -m68000 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>5200 : -m5200 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>5206e : -m5206e ; flags gcc CFLAGS <architecture>m68k/<instruction-set>528x : -m528x ; flags gcc CFLAGS <architecture>m68k/<instruction-set>5307 : -m5307 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>5407 : -m5407 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>cpu32 : -mcpu32 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>68040 : -m68040 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>68060 : -m68060 ; flags gcc CFLAGS <architecture>m68k/<instruction-set>soft-float : -msoft-float ; >So I've done a small patch to tools/build/v1/gcc-tools.jam adding >GCC_PREFIX variable, that if set up, will prefix names of all tools with >that prefix. > >So I can compile boost with > >bjam -sTOOLS=gcc -sGCC_PREFIX=arm-linux- stage > > Ah, your approach with using GCC_PREFIX is even better! Perhaps call it "-sCROSS_TARGET=arm-linux" ? >May be it will be useful to add the same functionality to >gcc-stlport-tools.jam or mingw. > >So please find attached changes. > >It is possible to include them into next boost release? > > I strongly vote for integrating cross-compilation support (at least using GNU toolchain, after all that's what most use!) in the Boost build system. But I don't know or use jam or bjam (other than building (parts of) boost itself), so unfortunately I would be of little help... :-( I replied to the post becuase somone else might find the information helpful as well, I hope. >By the way, the same problem appears when boost jam is used to compile >some third-party libraries. Is this place good to talk about boost jam >and its Jambase file? > > Probably move to the boost-build list? -wolfi -- DI Wolfram Stering (Entwicklung) HALE electronic GmbH., Salzburg, Austria Tel: +43 (662) 439011 550 Fax: +43 (662) 439011 9 Email: wolfram.stering(a)hale.at

2 1

How to create a smart pointer for a non-growable, size determined by run-time array
by yinglcs2＠yahoo.com 12 Jan '06

12 Jan '06

Hi, I have a non-growable, size determined by run-time array. For example convert this from Java to C: int[] func(int size) [ int[] array = new int[size]; return array; } I was told the only way I can do that is using STL vector. But if I need to have a smart pointer for this STL vector, should I use the c++ auto_ptr? or the Boost's shared_ptr with STL vector or Boost's shared_array? Thank you. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

3 3

callback and boost:function
by Lars Schouw 12 Jan '06

12 Jan '06

Guys, First thank you for a wonderful library and all the hours of work that is put into boost to make it what it is today. Is there anyway in C++ to setup a callback to a member function that takes two arguments I am trying to use boost::function but that only works with std::bind1st that does only take one parameter. I want to wrap some existing C code with C++. The function I want to call back is declareed like this: struct X { double foo(double x , void * p); }; Idealy I would like to use lambda to remove the void* and replace it with _1, _2 etc. But I am not sure how that works with callbacks. I had a look at the libsigc++ that seems to be able to handle callbacks with more arguments but the lambda stuff does not work under Visual Studio 2003 at least. Regards Lars Schouw --------------------------------- Yahoo! Photos Ring in the New Year with Photo Calendars. Add photos, events, holidays, whatever.

2 1

Re: [Boost-users] Question Regarding Boost 1.33.1 on MSVC.NET 2003
by Wu Yinghui, Freddie 12 Jan '06

12 Jan '06

Peter Dimov wrote: > The problem is that > > boost::bind(&OperatorMap::value_type::second, _1) > > returns an Operator by value (in 1.33.1) or by const reference (in the > current CVS). Since boost::bind needs to decide on a result type at bind > time, and since it doesn't know whether the argument coming in from the _1 > is const, it needs to conservatively assume const (Lambda doesn't have this > limitation because it deduces the return type at call time). > > Use > > boost::bind<Operator&>(&OperatorMap::value_type::second, _1) > > to tell boost::bind to return a non-const reference. Thanks for your quick reply! I noticed this problem at about the same time, and tried it out already. This modification indeed worked. Cheers, Freddie -- Wu Yinghui, Freddie Research & Development Software Engineer Volume Interactions Pte Ltd 1 Kim Seng Promenade, #12-01 Great World City East Tower Singapore 237994 Tel: +65 62226962 (Ext 216) Fax: +65 62226215 Email: yhwu(a)volumeinteractions.com URL: http://www.volumeinteractions.com Important: This message is intended for the recipient(s) addressed above. It contains privileged and confidential information. If you are not the intended recipient, please notify the sender immediately by replying to this message and then delete it from your system. You must not read, copy, use, or disseminate this communication in any form. Thank you.

1 0