April 2008 - Boost-users - lists.preview.boost.org

[Conversion] numeric_cast from floating to integral types
by Rainer Thaden 23 Apr '08

23 Apr '08

Hi all, I just discovered (by debugging instead of reading :-() that a numeric_cast from a floating to an integral type always truncates by default, that is, 26.99999 is casted to 26. Is there a way to tell numeric_cast to use the nearest integer? Of course I can add 0.5 to my value and cast afterwards but that's definitely ugly. Thanks in advance. Best regards, Rainer

3 4

[asio] obtaining a socket descriptor from ip::tcp::socket
by Matthias Vallentin 23 Apr '08

23 Apr '08

Hello, I would like to incorporate asio into my current project. However, I am restricted by a library that uses C-style socket descriptors to identify sockets, e.g. it contains functions similar to void foo(int fd, ...) Right now, I picture write an asio server that merely performs an asynchronous accept. Thereafter, the start() function in handle_accept() (see below) would invoke calls to the library such as foo(fd, ...) and spawn a dedicated thread per client that would gather data from the socket. void Server::handle_accept(const boost::system::error_code& e) { if (! e) { new_connection_->start(); new_connection_.reset(new Connection(...) //... In this example, asio would only be used to to handle the accept() and provide an outside interface to the rest of the application while the actual read()/write() calls are capsulated in the library. Is there a way to obtain a native socket descriptor from an ip::tcp::socket object? I am aware that mixing asio with C-style networking code totally breaks platform independence and it is generally a very ugly approach. On the other side, the C-style library would be capsulated behind one asio Server object, without interacting with the remaining application code. I appreciate any input on this issue, Matthias -- Matthias Vallentin vallentin(a)icsi.berkeley.edu pgp/gpg: 0x37F34C16

2 2

[iostreams] write function for sink
by Edward Diener 23 Apr '08

23 Apr '08

In iostreams I am creating a sink device and implementing the write function. In certain situations I do not want to write anything to the sink. In that case should I be passing back 0 as the streamsize or some other value ? BTW clicking on 'sink' in the individual classes list brings me to the page for 'device' not 'sink'.

1 0

Older Simpler Boost Library List
by Ray Burkholder 23 Apr '08

23 Apr '08

I see a current list of libraries at http://www.boost.org/doc/libs/1_35_0. On the old site, there was a nice table based grouping of libraries. I'm thinking it was easier to find something in the old table than on the new page. Does a 1.35.1 version of the old table style exist somewhere? -- Scanned for viruses and dangerous content at http://www.oneunified.net and is believed to be clean.

2 1

[serialization] Re-serialization ??
by Robert Bielik 23 Apr '08

23 Apr '08

Is there a way to do a "re-serialization" of an object structure? Let's say I have a couple of objects, and I serialize them to "State A", then change a few things. Ideally, I'd like to just deserialize "State A" so that I get the objects back to that state, without boost::serialization creating new objects for me. Is it doable? (it is for an undo implementation) TIA /Rob

5 4

Re: [Boost-users] [iterators] NULL pointer initialization and comparison using iterator_facade
by Lang Stefan 23 Apr '08

23 Apr '08

> On Tue, Apr 22, 2008 at 10:43 AM, Lang Stefan > <SLang(a)starragheckert.com> wrote: > > I have a big problem though: The project I am working on > is quite big > > and I intend to replace a particular typedef (currently > simply a pointer > > to a struct) with this iterator class. There are two kinds of > > instructions though that my iterator class currently can't > cope with: > > > > 1. Initializing a 'pointer' with NULL (in other words, I need a > > constructor with one argument of type int, but only in case that > > argument's value is 0) > > > > How about some really weird pointer type? If you did a member > function type that doesn't actually exist, you should be able to > construct with NULL, but nothing else. > Whatever you have in mind, I have no idea what you mean. Could you elaborate please? Btw., I did consider somehow introducing one of the boost smart pointer types to resolve this issue, but apparently even the smart pointers cannot directly deal with NULL values (e.g. comparison to NULL requires calling the get() method first). I could of course introduce special functions for semantically equal purposes, but that would force me to change several thousand lines of code to match the new syntax. I can think of better ways to waste my time... Is it possible to construct metafunctions that allow a function argument of NULL, but not any other integer value? Or at least cause the compiler to call out whenever a non-NULL argument is provided? Cheers, Stefan

2 1

[iostreams]combining serialization archive with iostreams filtering_streambuf
by cpper 23 Apr '08

23 Apr '08

Hi all: I have spent a day long in finding a way of combining serialization archive with iostreams filtering_streambuf but now I encounter a strange problem, here is my code: #include <sstream> #include <boost/iostreams/device/back_inserter.hpp> #include <boost/iostreams/filtering_stream.hpp> #include <boost/iostreams/copy.hpp> #include <boost/iostreams/filter/bzip2.hpp> #include <boost/archive/binary_oarchive.hpp> #include <boost/archive/binary_iarchive.hpp> #include <boost/serialization/string.hpp> #include <boost/iostreams/stream.hpp> #include <vector> int main () { std::vector<char> store; boost::iostreams::filtering_ostreambuf out; out.push(boost::iostreams::bzip2_compressor()); out.push(boost::iostreams::back_inserter(store)); { boost::archive::binary_oarchive oa(out); oa << "abc"; // archive closed when destructors are called //out.pop ();//call before oa destroyed, crashed!!!! } out.pop ();//call after oa destroyed the behavior seems normal //out.reset (); boost::iostreams::filtering_istreambuf in; in.push(boost::iostreams::bzip2_decompressor()); in.push(boost::make_iterator_range(store)); { std::string aft; boost::archive::binary_iarchive ia(in); ia >> aft; } } if I call pop() or reset() before archive destructor called. exception in: C:\Program Files\Microsoft Visual Studio 9.0\VC\crt\src\streambuf in this member method of std::streambuf int pubsync() error message is below: Unhandled exception at 0x10499231 (msvcp90d.dll) in basic_usage.exe: 0xC0000005: Access violation reading location 0xfeeeff1e. If I call out.pop() after oa destroyed,the behavior is exactly what I want. But I just wonder why it goes in this way? Code like below { boost::archive::binary_oarchive oa(out); oa << "abc"; } out.pop ();//call after oa destroyed the behavior seems normal which appears that I have done something wrong... I'm using VS2008TS under Win2003SP2 Can anybody help?

1 0

[serialization]
by Robert Bielik 23 Apr '08

23 Apr '08

I'm going to serialize my objects via the binary archives, since there seems to be some issues with the wide variants, I'd like to know what happens to wide strings serialized? Are they just copied verbatim, as wide char data that is? TIA /R

1 1

Boost::interprocess, managed_mapped_file not working properly on Mac
by Martin_Thomas＠McAfee.com 22 Apr '08

22 Apr '08

I was trying to make a small program that uses Multi-Index and Interprocess work on a Mac and was convinced that I was doing something wrong until I tried the simple example in the Multi-Index docs and found it also fails to work on Mac but works on Linux. The example is here: http://www.boost.org/doc/libs/1_35_0/libs/multi_index/example/ip_allocat or.cpp and the problem is that boost::interprocess::open_or_create does not work -it only creates. I built the example, add a few books and then quit. When I restarted the app on Linux, it listed the books from the previous run but the Mac version did not manage to do the same. Running on Linux Redhat FC8 and Mac OSX 10.5 Cheers // Martin

1 0

[extension] conflict with boost::optional
by David Matz 22 Apr '08

22 Apr '08

Hello, I have some trouble using boost::extension and boost::optional together in a project (MSVC 7.1). The compiler complains that boost::get is not an element of extensions::shared_library. Without the optional.hpp header everything works fine. It seems the for some reason the macro BOOST_EXTENSION_SHARED_LIBRARY_GET_FUNCTION isn't valid when shared_library.get<>() is called but I can't find out why. A small code snippet that reproduces the error file: test_extension_and_optional.cpp #include <boost/optional/optional.hpp> // included for some reason #include <boost/extension/shared_library.hpp> #include <boost/extension/convenience.hpp> // load_single_library void loadExtension () { boost::extensions::factory_map fm; // we don't need the following line to produce the compiler error //boost::extensions::load_single_library(fm, ".", "external_function_name"); } error: error C2039: 'boost::get': Ist kein Element von 'boost::extensions::shared_library' Thanks in advance. David -- David Matz ---------------------------- Software Developer Spatial View GmbH mail: david.matz(a)spatialview.com phone: +49 351 - 4 50 33 44 registration court: Amtsgericht Dresden, HRB 21104 managing directors: Wolfgang Opel, Jens Wiegrebe

3 8