February 2007 - Boost-users - lists.preview.boost.org

[asio] Using tcp::iostreams with Boost.Iostreams filtered_streambuf
by Darren Garvey 14 Feb '07

14 Feb '07

Hi, I was wondering about tcp::iostreams. I need to have the ability to write code like: obj<< "Hello world" and at intervals (of size) have a header attached and then have it written to an open socket. I have been looking into Boost.Iostreams to do this, but I was wondering if rather than implementing the whole streambuf/sink, I could just create a filtering_streambuf (naming convention taken from Boost.Iostreams) - which would attach the header when needed - and just instantiate a tcp::iostream object with that streambuf. I'm still very foggy on the Iostreams library - and streams/buffers in general - but does that sound like it could work? Sorry about the vagueness, Darren

2 1

[signals] rationale for signals::connection::disconnect() const
by Frank Mori Hess 14 Feb '07

14 Feb '07

I just noticed that boost.signals declares signals::connection::disconnect() as a const member function. Can anyone explain to me why that was done? It seems like a non-const operation to me. That's how I declared it in thread_safe_signals, and it doesn't seem to cause any problems there. -- Frank

1 0

How to block a signal
by Dahman 14 Feb '07

14 Feb '07

Hi, I have a signal with many slots connected to it and I have a class object witch want to have the possibility to block this signal at the start and unblock it later, but know nothing about what and how many slots are connected to it (then can't use connection.disconnect()). Is there any solution for this problem? Thanks for any help.

5 8

program_options boost::bad_any_cast: failed conversion using boost::any_cast
by Raiko Albrecht 14 Feb '07

14 Feb '07

Hi, since a while I always get this any-cast error when using program_options with type "std::string". to check it out I'd modified the simple example from boost for program_options a little bit (checking if any_type as it works - does, inserted a string-parameter). Compiling, starting with "--log astring" -> crash. starting without works at expected. (See appended source file) System is Debian etch, X86_64, gcc 4.1.2 20061115 boost: tried with debian package and self build variant. question now is: why? what may I do? As far as I remember on a unstable 32bit with gcc 4.0 it did work, but not sure if I got it running on a 64 bit system later. Hope for help Rajko

2 3

using unit_test_log
by Paez, Juan 14 Feb '07

14 Feb '07

Hello, I new using boost. I am trying to setup a set of test cases to test some api. I would like to be able to use the logging facilities that boost provides through the unit_test_log class. I understand that from the command line you can set the logging parameters. However, I want to be able to set those parameters from inside the my program executing the test cases. any ideas? Specifically, I want to be able to send the log output to specific files (the files change depending on the test suite I am running), the format, level, etc. Thanks, Juan

2 3

spirit and xml
by abir basak 13 Feb '07

13 Feb '07

Hi, I am at present using spirit for parsing unipen file as specified in http://www.unipen.org/dataformats.html . That is great and quite easy to parse, thanks to spirit. Now I am looking to use spirit for parsing an specific xml file ( w3c inkml file). So my intension is not to have a generic xml parser, rather than a specific xml parser (which also have some BNF grammar) . Anyone had used spirit for domain specific xml parsing? I believe using spirit will make it faster. Also I am interested to parse only a portion of the whole document at a time, and generate data from that portion only, rather generating data for whole DOM (The files are large, 4-20 MB typically) my xml file is something like, <page> <trace ref = "1"> 0,1,2,3,4,3,4,3,5,4,3,4 </trace> ... </page> note that inside <trace> the grammar is a BNF (comma sep float pairs mostly) Any comments, or snippets to show how to do it? -- Abir Basak, Member IEEE Software Engineer, Read Ink Technologies B. Tech, IIT Kharagpur email: abir(a)abirbasak.com homepage: www.abirbasak.com

5 7

boost::mutex question
by Anders Sundman 13 Feb '07

13 Feb '07

Hi all! I'm curious about the realization of the boost::mutex class (especially on W32). * How heavy is it to create a mutex object? * Are there any limits (OS recource bound or other) to how many mutex object that can be created? * How costly is it to 're-lock' a recursive mutex? Almost NOP or more costly? Are these questions (and other of the same type) answered somewhere in the docs? Finally, a more concrete question. Consider the following: A list of objects that needs to lock a mutex (all lock the same mutex). vector<CriticalWorker*> foo; ... for ( ... i = each foo ... ) i->doWorkThatLocks(); This code section will lock/unlock repeatedly. If I use a recursive mutex, will the following be much faster (since the locks in 'doWorkThatLocks' will be more light-weight)? lock(); for ( ... i = each foo ... ) i->doWorkThatLocks(); unlock(); Best Regards Anders Sundman

2 2

Preferred syntax on GCC 4.0
by Austin Shoemaker 12 Feb '07

12 Feb '07

I can't find this in the documentation or on the web, surprisingly, so I thought I'd ask here: Should I be able to use "preferred syntax" with boost::signal using GCC 4.0? I get "ISO C++ forbids delcaration of 'signal' with no type" errors, and portable syntax works fine. If it's relevant, I'm using the Apple build of GCC on OS X 10.4.8. Best, Austin

3 3

[lambda] std::string::find
by Bob Quampen 12 Feb '07

12 Feb '07

First, sorry for this total newbie question, i'm trying to do something like this: std::for_each ( std::istream_iterator<std::string>(ss), std::istream_iterator<std::string>(), boost::lambda::if_then ( !(boost::lambda::bind(&std::string::substr, boost::lambda::_1, 0, 1) == _ignore) && !boost::lambda::bind(&std::string::empty, boost::lambda::_1), (boost::lambda::bind(&std::string::find, boost::lambda::_1, _ignore), // compiler complains here ---------------------------------<snip>---------------------------------------------------------------------- i'm getting the following error: Error 23 error C2784: 'const boost::lambda::lambda_functor<boost::lambda::lambda_functor_base<boost::lambda::action<3,boost::lambda::function_action<3,T>>,detail::bind_tuple_mapper<const Arg1,const Arg2,const Arg3>::type>> boost::lambda::bind(const Arg1 &,const Arg2 &,const Arg3 &)' : could not deduce template argument for 'overloaded function type' from 'overloaded function type' am i not supplying the correct amount of parameters to the find method? -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

2 1

[OT] Is there a HTTP library in boost?
by Meryl Silverburgh 12 Feb '07

12 Feb '07

Hi, Is there a HTTP library in boost? I am looking for something which can handle http redirect. Thank you.

5 4

Boost::Date_Time - Incr and Decr Operators
by Steve Johns 12 Feb '07

12 Feb '07

Getting started with Boost::Date_Time ... It seems that this is OK: greg_year y1(2000); y1 = y1 + 1; but this is not: greg_year y1(2000); y1++; I take it that the increment and decrement operators are not implemented for greg_year (or the corresponding month, day, etc. types). Is this just incidental to the current implmentation, or is there some particular architectural or domain reason why ++ and -- aren't in the mix? - Steve

2 1

[signals] sandbox version status
by Timmo Stange 12 Feb '07

12 Feb '07

The modified version of Boost.Signals in the sandbox now compiles and passes the tests on my setup. Basic thread-safety is also in there, minus tracking and the combiner/signal-to-signal connection issues that have been discussed here recently. The thread-safe version will not support grouping and naming of slots, so there is a preprocessor flag "BOOST_SIGNALS_NO_LEGACY_SUPPORT" which changes the template signature of signalN. If defined, the Group and GroupCompare parameters are replaced with a ThreadingModel parameter (I chose to put it there and not after SlotFunction, but that may change). It defaults to signals::single_threaded which results in non-thread-safe behaviour. When signals::multi_threaded is given as an argument, the signal should be thread-safe. That is for now completely untested, though ;). As I was curious about the call performance, I ran a simple test, successively connecting up to 4 slots to the signal and calling it 10000 - 10000000 times. Here are the results: # Calls CVS SBL SBNS SBNM 1 10000 0.005 0.003 0.001 0.002 1 100000 0.057 0.037 0.019 0.025 1 1000000 0.582 0.362 0.209 0.269 1 10000000 5.882 3.659 2.073 2.668 2 10000 0.007 0.004 0.002 0.003 2 100000 0.075 0.044 0.027 0.031 2 1000000 0.769 0.461 0.269 0.322 2 10000000 7.668 4.535 2.744 3.274 3 10000 0.009 0.005 0.003 0.003 3 100000 0.095 0.054 0.035 0.038 3 1000000 0.964 0.555 0.360 0.411 3 10000000 9.643 5.538 3.525 4.010 4 10000 0.011 0.006 0.003 0.004 4 100000 0.114 0.061 0.038 0.044 4 1000000 1.155 0.641 0.419 0.451 4 10000000 11.482 6.378 4.124 4.519 CVS = The current CVS HEAD version. SBL = The sandbox version with legacy support (named slots). SBNS = The sandbox version without legacy support, single-threaded. SBNM = The sandbox version without legacy support, thread-safe. I used Visual C++ 2005 Express SP1 with /O2 optimization on an Athlon X2 3800+ for these tests. Results are in seconds. The comparison to the original CVS version is not quite fair, as I used dynamic linking there and static linking for the rest of the tests. I just did it to make sure my code was not significantly slower. The overhead for thread-safety does not grow with the number of connected slots, as I use a per-signal mutex and no per-slot locking. That looks good here, but I'm not yet sure it is the best solution in the long run. It would be nice to not block the entire signal for the whole time. As the new single-threaded implementation is somewhat faster than the one supporting named slots, I will add a signals::no_name type that can be used for the Group template argument when legacy support is enabled. That will result in the faster implementation being used without losing compatibility to existing code with grouped slots. Reports about problems with certain compilers and any kind of other input would be very appreciated. Don't judge too harshly on the code, I know it needs a lot of cleanup ;). Regards Timmo Stange

3 10

Performance of object_pool
by 蹬三轮的 12 Feb '07

12 Feb '07

The most important reason to use a memory pool is performance. BUT object_pool doesn't! When I use object_pool to construct a lot of objects, the destroy method become very slow. And I found object_pool use ordered_free to free the memory, and it amortized O(N). It is a very serious problem, I hope it can be fixed in the next release. thanks.

1 0

Compiling Example Code for Boost::Date_Time
by Steve Johns 11 Feb '07

11 Feb '07

Hi. I am new to Boost, and beginning with compiling example code for Boost::Date_Time. I've had success compiling the many examples given here: http://www.boost.org/doc/html/date_time/examples.html but the very first example given here: http://www.boost.org/doc/html/date_time/examples/general_usage_examples.html I cannot compile without errors. (Code given below) Changing week(1) to weeks(1) got me past the first hurdle, but still not home. The problem seems to have something to do with 'Sunday' not being of the correct type, but I'm not truly certain what is going on, hence this post ... As background, I'm using the MS VC++ 2005 compiler and Boost 1_33_1. This example fails compilation from both the IDE and the command line. I'd appreciate any insight into what would be needed to get this example to correctly compile and run. (And if there is a flaw in the example, I'd suggest that the docs be fixed to provide it in working form.) Thanks! - Steve __ using namespace boost::gregorian; date weekstart(2002,Feb,1); date weekend = weekstart + week(1); date d2 = d1 + days(5); date today = day_clock::local_day(); if (d2 >= today) {} //date comparison operators date_period thisWeek(d1,d2); if (thisWeek.contains(today)) {}//do something //iterate and print the week day_iterator itr(weekstart); while (itr <= weekend) { std::cout << (*itr) << std::endl; ++itr; } //input streaming std::stringstream ss("2004-Jan-1"); ss >> d3; //date generator functions date d5 = next_weekday(d4, Sunday); //calculate Sunday following d4 //US labor day is first Monday in Sept nth_day_of_the_week_in_month labor_day(nth_dow::first,Monday, Sep); //calculate a specific date for 2004 from functor date d6 = labor_day.get_date(2004);

2 1

[signals] thread-safe default combiner
by Frank Mori Hess 11 Feb '07

11 Feb '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'm noticing now that boost::last_value isn't thread-safe when used with a non-void return value, since a thread can't in general be sure the signal will have any slots connected to it when it is invoked. Would it be acceptable to make last_value throw an exception on the non-void, no slots case? Throwing an exception seems preferable to the dreaded "undefined behaviour". Or we could use a slightly modified default combiner with a different name. - -- Frank -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQFFzKtr5vihyNWuA4URAtTdAKCfJQicTdVtf5qaab5M2PCndy6bowCggwKi DnFSpe3zg5DJsKfCsfnOh7Q= =Tppf -----END PGP SIGNATURE-----

4 15

Does boost have something to support binary data streams?
by Michael Marcin 11 Feb '07

11 Feb '07

I commonly need to write and read binary data from files. The streams in the standard library are very tedious to use for such tasks. I'd like to be able to construct a stream thats loads a little endian binary data file and use it to write/read data with the insertion/extraction operators on big endian or little endian systems. Does anything like this exists? Thanks, Michael Marcin

5 8

Error in compiling Boost thread example
by Meryl Silverburgh 11 Feb '07

11 Feb '07

Hi, I am trying to compile the boost thread that I found here: http://www-eleves-isia.cma.fr/documentation/BoostDoc/boost_1_29_0/libs/thre… But i get the following error: Can you please tell me what do i need to do to fix it? ./thread.o: In function `main':../thread.cpp:27: undefined reference to `boost::thread::thread(boost::function0<void, std::allocator<boost::function_base> > const&)' :../thread.cpp:28: undefined reference to `boost::thread::join()' :../thread.cpp:28: undefined reference to `boost::thread::~thread()' :../thread.cpp:28: undefined reference to `boost::thread::~thread()' ./thread.o: In function `thread_alarm::operator()()':../thread.cpp:11: undefined reference to `boost::xtime_get(boost::xtime*, int)' :../thread.cpp:14: undefined reference to `boost::thread::sleep(boost::xtime const&)' collect2: ld returned 1 exit status make: *** [sandbox] Error 1 make: Target `all' not remade because of errors. Build complete for project sandbox

3 7

how to use boost to bind to an attribute?
by Meryl Silverburgh 10 Feb '07

10 Feb '07

Hi, I have code like this which sort a Rect, which compare value of getY(), like this: vector<Rect*> rl; ::sort(rl.begin(), rl.end(), bind(less < int >(), bind(&Rect::getY, _1), bind(&Rect::getY, _2))); how can I change it so that it compares with the 'y' attribute? like, rect1.y < rect2.y? Thank you.

3 4

[property_map] bit vector as property map
by Sebastian Weber 10 Feb '07

10 Feb '07

Hello everybody! I'm using the property_map scheme to pass some data-structures to vistors of BGL-algorithms. So far I usually use a vector in the manner: std::vector<std::size_t> props(num_vertices(my_graph), 0) Then I do something like bfs_search(..., make_my_vistor(&props[0]), ...) which works ok and passes me a correct property_map into the visitor created by make_my_visitor. Within that visitor I can then do the usual stuff like put(the_props_map, vertex_id, someValue) At least that's how I understood, how the property_map concept works. However, if I use a bit-vector the above scheme breaks down to work: std::vector<bool> bit_props(num_vertices(my_graph), false) bfs_search(..., make_my_vistor(&bit_props[0]), ...) The compiler compiles these lines, but complains that a temprorary is passed as a reference to make_my_visitor and the code generated crashes. My system is a Ubununtu Edgy with boost 1.33.1 and g++ 4.1.2. Any ideas why this happens and what I can do here? Thanks a lot in advance. Greetings, Sebastian

3 4

FW: Question about Boost.Test
by Ranji Raghavan 09 Feb '07

09 Feb '07

-----Original Message----- From: Ranji Raghavan [mailto:ranji.raghavan@openspirit.com] Sent: Friday, February 09, 2007 11:09 AM To: ''boost-users(a)lists.boost.org' Subject: FW: [Boost-users] Question about Boost.Test I am rookie as far as BOOST.TEST is concerned - I am trying to compile and link a simple BOOST.TEST program using VS 2005 (SEE EMAIL BELOW FOR STEPS). When I try and this simple plagiarized test program listed below, I get a bunch of LINKER errors - Error 1 error LNK2019: unresolved external symbol "void __cdecl boost::test_tools::tt_detail::check_impl(class boost::test_tools::predicate_result const &,class boost::basic_wrap_stringstream<char> &,class boost::unit_test::basic_cstring<char const >,unsigned int,enum boost::test_tools::tt_detail::tool_level,enum boost::test_tools::tt_detail::check_type,unsigned int,...)" (?check_impl@tt_detail@test_tools@boost@@YAXABVpredicate_result@23@AAV?$basi c_wrap_stringstream@D@3@V?$basic_cstring@$$CBD@unit_test@3@IW4tool_level@123 @W4check_type@123@IZZ) referenced in function "void __cdecl free_test_function(void)" (?free_test_function@@YAXXZ) NewTest.obj Error 2 error LNK2019: unresolved external symbol "public: void __thiscall boost::unit_test::test_suite::add(class boost::unit_test::test_unit *,unsigned long,unsigned int)" (?add@test_suite@unit_test@boost@@QAEXPAVtest_unit@23@KI@Z) referenced in function "class boost::unit_test::test_suite * __cdecl init_unit_test_suite(int,char * * const)" (?init_unit_test_suite@@YAPAVtest_suite@unit_test@boost@@HQAPAD@Z) NewTest.obj Etc etc ... May be I am missing something trivial that is causing this problem. Appreciate any help rendered.... Ranji -----Original Message----- From: Ranji Raghavan [mailto:ranji.raghavan@openspirit.com] Sent: Thursday, February 08, 2007 9:51 PM To: 'boost-users(a)lists.boost.org' Subject: RE: [Boost-users] Question about Boost.Test I have downloaded and built the BOOST libraries using VS2005 - I have pointed the VC++ directories INCLUDE files to the BOOST install directory and the Library files to the folder where the BOOST libraries were built to. But when I try even the simplest of test programs from one of the examples you had written - see below #include <boost/test/unit_test.hpp> using boost::unit_test_framework::test_suite; // most frequently you implement test cases as a free functions void free_test_function() { // reports 'error in "free_test_function": test 2 == 1 failed' BOOST_CHECK(2 == 1); // non-critical test => continue after failure } test_suite* init_unit_test_suite( int, char**) { test_suite* test= BOOST_TEST_SUITE( "Unit test example 1" ); // this example will pass cause we know ahead of time number of expected failures test->add( BOOST_TEST_CASE( &free_test_function ), 1 /* expected one error */ ); return test; } I receive LINK errors - any suggestions ? -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Gennadiy Rozental Sent: Wednesday, February 07, 2007 4:09 PM To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] Question about Boost.Test "Ranji Raghavan" <ranji.raghavan(a)openspirit.com> wrote in message news:000a01c74a28$df145f30$e90a0a0a@openspirit.com... > Actually, I am just looking into BOOST reading the documentation etc > Do you recommend any particular version of BOOST that I should install and > play with? I recommend you to use the latest version. Try 1.34 RC. > I am looking to define all the test data in XML files - maybe > 1000's of them - have a class that parses this XML file (some tests have > about 7-8 pieces of disparate primitive data types string, double, int > etc)and pass the array of class instances to a test method for repetitive > testing. If this is not pragmatic, could you recommend another way? Use BOOST_PARAM_TEST_CASE. It works differently in latest version then before. You will have less problems with parametrs object lifetime management. Let me know if you have any problems. Gennadiy _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

serialization stream error
by pwm 09 Feb '07

09 Feb '07

I am trying to use the boost serialization library to save an objects from a class heiarchy. I keep running into a stream error when deserializing a pointer from an xml archive. I have everything registered (that took a while, but I have that problem figured out, I think), and the serialization to an xml archive seems to work just fine. I have stepped through the deep nesting of calls to the exception and it is being thrown because the stream says it is bad when trying to read the end tag. I also tried it with text archives, and I get the same problem. I have done the obvious (to me at least) things such as class export/registration, us std::ios::binary for output/input stream instances. Any advice on how to track down this problem with a stream error? I have gone back to first principles with a simple example and that all seems to work. So it is obviously user error. I am trying to understand what is wrong with my more involved case. I appreciate any advice someone could give me.) Thanks, Phil

3 7

[Python] Bug when exporting function overloaded on base and derived classes
by Edson Tadeu 09 Feb '07

09 Feb '07

Hi, I've found a bug where Boost.Python calls the wrong function overload. Here is a small unit test: ===== The C++ module "_test.cpp": ===== #include <boost/python.hpp> #include <string> class Base { }; class Derived : public Base { }; std::string f(Base& b) { return "Base"; } std::string f(Derived& d) { return "Derived"; } BOOST_PYTHON_MODULE( _test ) { using namespace boost::python; class_<Base>("Base"); class_<Derived, bases<Base> >("Derived"); def("f", static_cast<std::string (*)(Derived&)>(&f)); def("f", static_cast<std::string (*)(Base&)>(&f)); } ===== The small test in python: ===== >>> from _test import * >>> f(Base()) 'Base' >>> f(Derived()) 'Base' ===== Interestingly enough, if the function overloads are exported in the inverse order, it actually works, i.e., if using this code: ===== BOOST_PYTHON_MODULE( _test ) { using namespace boost::python; class_<Base>("Base"); class_<Derived, bases<Base> >("Derived"); def("f", static_cast<std::string (*)(Base&)>(&f)); def("f", static_cast<std::string (*)(Derived&)>(&f)); } ===== I do not know very well how does the boost.python registry works, but it seems that it is selecting the first match instead of testing all matches and selecting the "best one".

1 0

[asio].[hp-ux].[1_33_1] Build related issue
by Gaurav.Jain＠iflexsolutions.com 09 Feb '07

09 Feb '07

Hi, I have tried using boost 1_33_1 on HP UX machine, while compiling the program I am getting some errors. Can anybody help me out? The detail of the compilation is as follows: Boost Version: 1_33_1 bjam install -sTOOLS=acc --prefix="/home/pthakre/boost" --builddir="/tmp" \ "-sBUILD=release <runtime-link>dynamic/static <threading>multi <cxxflags>+DD64 <cflags>+DD64 <linkflags>+DD64" --without-python Jamfile Compiler Details: aCC: HP C/aC++ B3910B A.06.13 [Nov 27 2006] OS Details: HP-UX B.11.23 ia64 While compiling "daytime_server.cpp" with -D_HPUX_SOURCE -D_XOPEN_SOURCE_EXTENDED, I am getting the following error: "boost-1_33_1/include/boost/asio/detail/socket_types.hpp", line 136: error #2020: identifier "ip_mreq" is undefined typedef ip_mreq in4_mreq_type; Also, from /usr/include/netinet/in.h #ifndef _XOPEN_SOURCE_EXTENDED /* * Argument structure for IP_ADD_MEMBERSHIP and IP_DROP_MEMBERSHIP. */ struct ip_mreq { struct in_addr imr_multiaddr; /* IP multicast address of group */ struct in_addr imr_interface; /* local IP address of interface */ }; #endif /* ! _XOPEN_SOURCE_EXTENDED */ Compiling "daytime_server.cpp" without -D_HPUX_SOURCE -D_XOPEN_SOURCE_EXTENDED, I m getting 22 errors. I have done the following changes: Line 575 ./asio/detail/socket_ops.hpp: #if defined(BOOST_WINDOWS) || defined(__CYGWIN__) || defined(__hpux) Line 611 ./asio/detail/socket_ops.hpp: #if defined(BOOST_WINDOWS) || defined(__CYGWIN__) || defined(__hpux) Line 274 ./asio/error.hpp: #if defined(__sun) || defined(__QNX__) || defined(__hpux) Line 109 ./asio/system_exception.hpp: #elif defined(__sun) || defined(__QNX__) || defined(__hpux) Regards, Gaurav

3 7

easy serialization question
by Mark Wyszomierski 09 Feb '07

09 Feb '07

Hi, I'm using the boost serialization methods, they're great. I'm using some 3rd party classes that I don't have access to, so I can't add a serialize() member function to them. It's something like: class CMyClass { int m_nWhatever; CSomeClass m_sc; // I am just using this from someone else's library.... void serialize(Archive &ar, const unsigned int version) { ar & version; ar & m_nWhatever; ar & m_sc; // fails!!! } } how can I get around this? Thanks, Mark

5 8

[signals] thread-safe combiner()
by Frank Mori Hess 09 Feb '07

09 Feb '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 In implementing some of the missing parts of thread_safe_signals, I noticed that boost::signal provides a non-const combiner() function. In it's current form, it can't be made thread-safe against signal invocation. Is it worth changing the prototype from combiner_type& combiner(); to void set_combiner(const combiner_type &); so the combiner can only be modified while holding the appropriate lock? - -- Frank -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQFFy55q5vihyNWuA4URAiHbAJ45kSHJDeBOCKy+s5Yc6TwzqoclkACgx7PY vLAMNIfrlNQWu1jXFppxOj4= =9yLn -----END PGP SIGNATURE-----

3 2