- Boost-users - lists.boost.org

wrong extension for shared libraries on HP/UX
by Oliver Kowalke 20 Apr '05

20 Apr '05

Hello, I've compiled boost on HP-UX with gcc-3.4.2. The shared libraries ends with *.so. On HP/UX the extension for shared libraries should be *.sl. So - is it a bug or what went wrong? thx, Oliver -- +++ GMX - Die erste Adresse f�r Mail, Message, More +++ 1 GB Mailbox bereits in GMX FreeMail http://www.gmx.net/de/go/mail

1 0

wrapping an mmap'ed pointer
by Davi de Castro Reis 20 Apr '05

20 Apr '05

Hi all, I am writing a class that returns some pointer to buffers in a mmaped area. Since this buffers have constant size, I am now wrapping them in a std::pair<const uint32, char *buf>. This works fine, but I can bet that boost have something better, with correct semantics for operator * and, maybe, ownership semantics. Another problem that I have with std::pair<const uint32, char *buf> is that I don't get the correct cv-qualified semantics with a const std::pair<const uint32, char *buf>. I would like the const modified version of my type to behave like a const std::pair<const uint32, const char * const buf>. The reason for this is that I will, then, be able to receive const char * buffers in my input. Check sample code below: class A { typedef std::pair<const uint32, char *> value_type; void put(const value_type buf); value_type get(); const value_type get() const; } Buffers in the input (put method params) will be copied inside my data structure and their memory management is up to the user. Buffers in the output (get method return values) are pointers to a mmap area and should not be freed. So, if I do: A a; a.put(make_pair(strlen("test"), "test")); Code doesn't compile because cv-qualification of the char * member of buf does not get correct. The only way to get out is to write this rather ugly class A. class A { typedef std::pair<const uint32, char *> value_type; typedef const std::pair<const uint32, const char * const> const_value_type; void put(const_value_type buf); value_type get(); const_value_type get() const; } Is there a better solution for this? boost::array doesn't fit the job because the size parameter is compile time. Smart pointers, besides missing the size member, calls delete() when it is not desired. So, a class to fit the job needs to have the size member (if there is no such class, it is ok to stick to std::pair or boost::tuple), and expected cv-qualification for the char *, meaning that neither the pointer nor the pointed-by area can be modified (btw, this problem arises often when I am working with boost::mpl, so a help here will be greatly appreciated). Thanks in advance, []s Davi de Castro Reis

1 0

aliasing make_tuple
by Davi de Castro Reis 20 Apr '05

20 Apr '05

Hi, I have a class where I use a tuple as input for some of my methods. The reason for this is that the number of parameters for these methods is a compile-time variable. See sample code below. template <uchar nparams, typename T1, typename T2 = none, typename T3 = none ...> class A { void f(boost::tuple<char, char, ..., char>); } So users can do something like A a<3, char, char, char> a.f(make_tuple('a', 'b', 'c'); The problem is that make_tuple is not a nice name, since it does not reflect the semantics of the parameter. Suppose I would like to call it key, so I can call A a<3, char, char, char>; a.f(key('a', 'b', 'c')); So, in my context, everything gets much more clear. Does someone know an elegant way of doing this? As I can thing, the only way I can do it, is to create a templatized free function called key, for which I will need to write 10 overloads (the max number of nparams, and also boost tuple, if I am not mistaken). Besides being a rather ugly solution, I will need to pay two copies of my objects before sending them to the f() method (with make_tuple I pay only one copy, which is also suboptimal, but, anyways...). Using #define is not an option since key is a very common name (the class is in a namespace). Ugly names such as MY_LIBRARY_KEY() are not a very nice also. Well, any tips? Solutions involving boost::mpl would be the best, but I couldn't find none. []s Davi de Castro Reis

2 3

RE: [Boost-users] Re: why so many .so and .a files for boost?
by Sliwa, Przemyslaw (London) 20 Apr '05

20 Apr '05

I was wondering why I get duplicates on my Linux box as well. The files are with 1_32 and without extension and they are exactly the same. Full debug/release install takes about 0.5G. This is a lot!! Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Russell Hind Sent: 20 April 2005 09:09 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: why so many .so and .a files for boost? Rene Rivera wrote: > >> libboost_signals-gcc-1_32.a >> libboost_signals-gcc-1_32.so >> libboost_signals-gcc.a >> libboost_signals-gcc.so > > 'mutil-thread' and 'sigal-thread' support? What's the relationship > between these versions and the compiler options like '-pthread' ? > > Answers at.. http://www.boost.org/more/getting_started.html#Results But why duplicates of each? On windows I get libboost_signals-bcb-1_32.dll and libboost_signals-bcb.dll Which are both the same. Why are there libraries without the version number in created? Cheers Russell _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

3 2

why so many .so and .a files for boost?
by Tom Tan (SH/RDC) 20 Apr '05

20 Apr '05

Hi, all, Why are there so many .so and .a files in $BOOT_HOME/lib? Take boost_singal as an example, there are about a dozen . They are: libboost_signals-gcc-1_32.a libboost_signals-gcc-1_32.so libboost_signals-gcc.a libboost_signals-gcc.so libboost_signals-gcc-d-1_32.a libboost_signals-gcc-d-1_32.so libboost_signals-gcc-mt.a libboost_signals-gcc-mt-d-1_32.so .... I wonder what are 'd' and 'mt' mean? "debug" and "mutil-thread"? If so, i wonder when i use a 'mt' thread , e.g. libboost_signals-gcc-mt-d-1_32.so, will i have to use -pthread for Linux and -pthreads for Solaris along with my other g++ compile options. Why? Another, for stdc++.so, are there two versions , each for 'mutil-thread' and 'sigal-thread' support? What's the relationship between these versions and the compiler options like '-pthread' ? BR. /TomTan

3 2

Tokenizer in boost 1.32 adds more requirements on Type template parameter for Tokenizer class
by Sebastien Martel 20 Apr '05

20 Apr '05

Hello, It seems that the optimization introduced in token_functions.hpp consisting of converting += into assign calls introduced a new requirement on Type : template < typename TokenizerFunc = char_delimiters_separator<char>, typename Iterator = std::string::const_iterator, typename Type = std::string > class tokenizer { //... } In my case, Type is a custom string class that doesn't implement assign, but has operator +=. I want to avoid changing the interface of the string class. Is Tokenizer reasonable in requiring Type to implement assign ? Could assign_or_plus_equal (token_functions.hpp) fall back on += if assign cannot be found ? It currently does so if the iterator is an input iterator, how about expanding it to test Type traits(!) for an assign member ? Ideas ? Thanks! -- Best regards, Sebastien mailto:smartel@real.com

1 0

Passing BOOST arrays over CORBA
by Simon Monckton 19 Apr '05

19 Apr '05

Hello all: Do any of you have experience passing BOOST vectors/arrays over CORBA? Short of writing a BOOST idl, is there a way of getting at and passing raw array data? -- --------------------------------------------------------------------- Dr. Simon Monckton Defence Scientist Tactical Vehicle Systems Section DRDC Suffield Defence Research and Development Canada Phone: (403) 544-4724 Email: Simon.Monckton(a)drdc-rddc.gc.ca ---------------------------------------------------------------------

3 4

Re: [Boost-users] Re: How to serialize a vector?
by kauai 19 Apr '05

19 Apr '05

Thank you, actually was a newbie mistake. Forgot to include: #include <boost/serialization/vector.hpp> --- Jeff Flinn <TriumphSprint2000(a)hotmail.com> wrote: > > "kauai" <kauai_diver(a)yahoo.com> wrote in message > news:20050419013506.49781.qmail@web60707.mail.yahoo.com... > > Hi all, > > > > I am having problems serializing a vector. > > > > When I have code like this: > > > > ar & m_attrVec; // in my serialize function > > > > I get this error: > > > > xyz.cpp:213: no match for > > `boost::archive::text_iarchive& << > > std::vector<Attribute*, std::allocator<Attribute*> > >&' > > operator > > > > Am I missing something, > > Have you read > http://www.boost.org/libs/serialization/doc/index.html. > > The error is most likely related to your Attribute > pointer, and not so much > with the vector itself. > > Jeff Flinn > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > __________________________________ Do you Yahoo!? Plan great trips with Yahoo! Travel: Now over 17,000 guides! http://travel.yahoo.com/p-travelguide

1 0

Filesystem woes
by steve ahlgren 19 Apr '05

19 Apr '05

Hello all : I'm having great difficulty with boost::filesystem::path (again) and wondering if anyone else has seen this behavior. I'm using vc7.1 (msvc 2003) testing with both static and dynamic link boost::filesystem libraries. The behavior is as follows : 1. The boost regression tests all pass ( ie, in BOOST_ROOT/status running "bjam --dump-tests" and sorting through the output ). 2. If I use fs::path in my code with an absolute windows path ( ie c:/temp ) it _always_ throws. As an example, when I include the following code from the fs test file path_test.cpp in my own code it fails : fs::path p = fs::path( "c:/", fs::native ); assert( p.relative_path().string() == "" ); assert( p.branch_path().string() == "c:" ); assert( p.leaf() == "/" ); assert( p.root_name() == "c:" ); assert( p.root_directory() == "/" ); assert( p.root_path().string() == "c:/" ); assert( p.has_root_path() ); assert( p.has_root_name() ); assert( p.has_root_directory() ); assert( !p.has_relative_path() ); assert( p.has_leaf() ); assert( p.has_branch_path() ); assert( p.is_complete() ); 3. The exceptions are being chucked by the checker code; if I use boost::filesystem::no_check it doesn't fail. The checker complains that "c:" is invalid but clearly it's not ! 4. Before anyone mentions it, the rtls are consistent across all libraries including boost. Perhaps that's the problem though? Do people use boost successfully with /MD and /MDd? Has anyone else seen similar behavior or have any suggestions about what may be happening? As a long shot is this something to do with wchar_t settings on msvc ( /Zc:wchar_t )?? If it's not too much trouble (and I'm assuming it doesn't break the license ) can someone send me the debug build of boost::fs lib (either static or import/dll)? I'll happily pay you back in good kharma! Many thanks in advance; Steve a.

3 3

How to serialize a vector?
by kauai 19 Apr '05

19 Apr '05

Hi all, I am having problems serializing a vector. When I have code like this: ar & m_attrVec; // in my serialize function I get this error: xyz.cpp:213: no match for `boost::archive::text_iarchive& << std::vector<Attribute*, std::allocator<Attribute*> >&' operator Am I missing something, or do I have to use a different method to serialize vectors? Thanks for any help, Kauai __________________________________ Do you Yahoo!? Plan great trips with Yahoo! Travel: Now over 17,000 guides! http://travel.yahoo.com/p-travelguide

2 1