- Boost-users - lists.boost.org

RE: [Boost-users] Re: Boost problems
by Puverle, Tomas (IT) 18 Sep '04

18 Sep '04

> That's been cleaned up in the latest CVS. Great! So that's 2 problems down, 2 to go. Do you think you could comment on the other 2 unresolved issues from my original email? (the defines in regex and Boost.build problem/feature request?) Thanks a lot. Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

2 1

Boost problems
by Tom Puverle 18 Sep '04

18 Sep '04

Hey boosters, We have found some problems with the following files: boost_1_31_0/libs/regex/src/cpp_regex_traits.cpp boost_1_31_0/libs/test/src/test_tools.cpp They use definitions from <ios> but never actually include it - they are relying on the fact that in some STLs the headers are pulled in by other header files. Unfortunately, under our STL (RW STL 6.0) this is not the case and hence the files fail to compile. Add the following include line to both to fix the problem: #include <ios> There is another STL related problem in boost_1_31_0/boost/regex/config/cwchar.hpp which manipulates some macros such as iswalnum etc. These are originally pulled in from <wctype.h> and <wchar.h> under /usr/include. With gcc, <cwctype> and <cwchar> undefine these macros as the first thing, however, our STL's headers don't. That leads to compilation errors from the above header. To fix this, add #undef iswalnum #undef iswalpha #undef iswblank #undef iswcntrl #undef iswdigit #undef iswgraph #undef iswlower #undef iswprint #undef iswprint #undef iswpunct #undef iswspace #undef iswupper #undef iswxdigit #undef iswctype #undef towlower #undef towupper #undef towctrans #undef wctrans #undef wctype when RW STL 6.0 is being used. The final problem is more of a feature request: In our environment, we build all shared objects with all external references unresolved, so that a .so A never depends on a shared library B. The symbols get only resolved once the final executable is linked. This causes us problems with boost when building the regression tests, as the final Jam link rule adds all additional libraries *before* the final libboost_*. That way, we end up with unresolved symbols to e.g. our STL. Is there any way in which this requirement could be incorporated into Boost.Build? I guess a simple solution would be to have another variable called e.g. ADDITIONAL_LIBS_LAST which would be passed as the last thing on linker line in the link rule. HW & SW details: OS: Red Hat Enterprise Linux 2.1 & 3.0 HW: various Compiler: gcc 3.2.3-20 STL: RogueWave 6.0 Boost: 1.31.0 Thanks, Tom _________________________________________________________________ It's fast, it's easy and it's free. Get MSN Messenger today! http://www.msn.co.uk/messenger

2 1

BOOST 1.32 soon?
by Daniel L Elliott 18 Sep '04

18 Sep '04

Hello, Will there be a BOOST 1.32 tagged soon? Thank you, to everyone, for this contribution! -- dan ******************* Dan Elliott *******************

4 7

Can boost::tokenizer tokenize 2byte character string?
by Lee, Joo-Young 18 Sep '04

18 Sep '04

Hi. I try to use 'boost::tokenizer<boost::char_separator<char> >' to separate 2byte character string like Korean, Japanese, or Chinese. But, I found that it does not works correctly. Is there a solution? Thanks for the help, Lee Joo-Young

1 0

Re: Newbie help in compiling boost
by Jeff Flinn 17 Sep '04

17 Sep '04

"David Abrahams" <dave(a)boost-consulting.com> wrote in message news:uzn3p2g31.fsf__38100.3229799264$1095428766$gmane$org@boost-consulting.com... > Murali Krishnan Ganapathy <gmkrishn(a)cs.uchicago.edu> writes: > > > When boost.jam tries to make a directory it tries to access %TEMP%. > > Machine complains that there is no "C;\DOCUME~1\...." (it is not > > checking the 8.3 version I guess) > > and the mkdir1 fails. Is the ';' here a typo, or is that what TEMP and TMP are really defined as? This is the 8.3 version, so I don't understand what your saying here. > > The work around is to have a simple %TEMP% and %TMP%. > > I reset %TEMP% and %TMP% to "C:\Temp" (no spaces and less than 8 > > chars), now boost is > > building as I write this. > > > > I guess this should be considered as a minor bug in bjam? > > Maybe so. Any Win/DOS experts out there who can suggest a correct > fix? NT, 2K and XP maintain "C:\Documents and Settings\<user-name>\Local Settings\Temp" directories for each user. Changing this to C:\Temp could cause problems for users w/out admin privileges. I've no problems with vc7.1 builds, and I imagine a lot of others utilize the same platform and can build as well. So it's not a general problem. The possible sources of this problem could be: - some odd access settings on the directory ( can even occur with admin privilege) - a lot of files in the temp directory from other misbehaving app's - environment variable definitions were mucked up by manual editing - generally mucking about with these dir's by the user can cause problems Jeff F

5 4

RE: [Boost-users] Boost problems
by Puverle, Tomas (IT) 17 Sep '04

17 Sep '04

On a similar note, I also noticed the following: iterator/detail/facade_iterator_category.hpp:# include <boost/python/detail/indirect_traits.hpp> iterator/indirect_iterator.hpp:#include <boost/python/detail/indirect_traits.hpp> It doesn't seem right that iterator should be including headers from python/detail. Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

2 1

boost.mem_fn with _com_ptr_t
by Andy Tompkins 17 Sep '04

17 Sep '04

Hi, I am trying to use com smart pointers with boost::mem_fn I defined the following so that mem_fn would know about _com_ptr_t s: template <typename Interface> inline typename _com_ptr_t<Interface>::Interface* get_pointer(_com_ptr_t<Interface> const& p) { return p.GetInterfacePtr(); } but when I use the function object returned from boost::bind, I get a null pointer in my _com_ptr_t ADODB::_RecordsetPtr recordSet(__uuidof(ADODB::Recordset)); ... boost::bind(&ADODB::_Recordset::Close, recordSet)(); The problem seems to be in mem_fn_template.hpp The function class BOOST_MEM_FN_NAME(mf0)::operator()(U &u) const calls BOOST_MEM_FN_RETURN call(u, &u); and _com_ptr_t::operator&() calls _Release() and sets m_pInterface = NULL; Is there any way to get this to work correctly? Thanks in advance. Andy.

1 0

Re: [Boost-users] sort implementation based on swap
by John Meinel 17 Sep '04

17 Sep '04

Howard wrote: > > On Sep 15, 2004, at 5:28 PM, Russell Mok wrote: > >> I know this is kinda off topic. >> >> I am trying to sort some objects which are expensive to copy. Does >> anyone know a sort implementation that uses neither copier nor >> assignment? The STLport version I have is using both. Worse, some >> routines are passing arguments as T instead of const T &. I did >> search boost and google but I couldn't find what I want. > > > The Metrowerks implementation of std::sort uses swap (unqualified) > exclusively. > > -Howard > > Another idea which I think works out for simplicity is to just sort an array of pointers, and then once all items are in order, you change the ordering on disk. I think this would be your biggest performance gains, as most sorting has a lot of intermediate steps. You can do all those fast with the pointers, and then do the slow sort of the files. John =:->

1 0

Newbie help in compiling boost
by Murali Krishnan Ganapathy 17 Sep '04

17 Sep '04

Platform: Windows XP Compiler: VC7 (Visual Studio .NET) Shell: Cmd.exe BJAM: Prebuilt version 3.1.10-1-ntx86 from sourceforge Boost Version: 1.31.0 On the prompt: * set PYTHON_ROOT=C:\Python23 * set PYTHON_VERSION=2.3 * C:\Program Files\...\VC7\bin\VCVARS32.BAT (execute this) * bjam "-sTOOLS=vc7" install I get a whole lot of lines of the form ...skipped XXX for lack of YYY... and finally get ...failed updating 2 targets ...skipped 3077 targets Any pointers on what I am doing wrong. On the prompt I am able to run "cl, link, nmake" without any problems (i.e. path is set correctly by vcvars32.bat). - Murali

3 4

RE: [Boost-users] Boost problems
by Puverle, Tomas (IT) 17 Sep '04

17 Sep '04

> > We have found some problems with the following files: > > > > boost_1_31_0/libs/regex/src/cpp_regex_traits.cpp > > boost_1_31_0/libs/test/src/test_tools.cpp > > > > They use definitions from <ios> but never actually include it > > - they are > > relying on the fact that in some STLs the headers are pulled in by > > other header files. Unfortunately, under our STL (RW STL > 6.0) this is > > not the case and hence the files fail to compile. Add the > following > > include line to both to fix the problem: > > > > #include <ios> > > Ok. > > Gennadiy. Gennadiy, have you any news about the second problem also related to regex, as described in my original email? Thanks, Tom -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

1 0