- Boost - lists.boost.org

[interprocess] shared memory lifetime
by Kim Barrett 11 Mar '06

11 Mar '06

In reply to my review comments, Ion Gaztanaga wrote 2/17/06: > > A reference count of create + open - close is maintained for shared > > memory objects. If this count reaches zero, the shared memory > > object is unlinked (at least in the posix version). The lack of any > > documented unlink mechanism might lead one to guess that something > > like this is going on, and is documented by the last paragraph of > > this section ("When all processes ... close ..., the shared memory > > segment is destroyed"). A bit more emphasis might be useful here. > > On the other hand, I'm actually not convinced this is a good idea. > > It is certainly fragile, in that a program which crashes (for > > whatever reason) won't close any shared memory objects it has open, > > resulting in the reference tracking getting messed up. > > You can have problems in POSIX systems, but I couldn't find a better way > to implement this. If a program crashes there is no way control > anything. The unique solution would be to provide a function that > destroys all named objects that I can register with every creation so > that you can catch signals and call that functions to free all objects. > In windows the OS frees the resources automatically. For standard C++ > IPC mechanisms I would request OS help for program crashes, just like > heap memory is freed automatically. > > > For example, one couldn't start up a > > process which parses some data into an in-memory format that it > > records in shared memory and then exits, with other programs saving > > parsing time by just getting the information from shared memory. > > This doesn't work if those other programs don't get around to > > opening the shared memory before the parser program exits. > > I would like to implement the POSIX-like behaviour in windows, but that > would require some permanent store that or a server/ process/service > that serves named IPC mechanisms windows. You can use memory mapped > files for this behaviour. Take in care that POSIX unlink mechanism is > also complex so that if a process unlinks the shared memory, if another > process can create a new shared memory with the old name while older > processes are still attached to the old shared memory. I need help from > POSIX experts. First let me make sure I understand what is going on here. On Windows, the shmem library is presently using create_file_mapping and open_file_mapping, while on POSIX systems it is using shm_open. An object created with create_file_mapping exists until there are no references, assuming I'm understanding what you've said. (I don't have (easy) access to Windows API documentation, so can't go look up this information. Please correct me if you see any confusion.) A POSIX shared memory object exists from the time it is created until it is unlinked and all references are gone (or the system is rebooted). (It becomes inaccessible to further shm_open calls if unlinked, but remains open to processes that had already opened it.) What the present library implementation is trying to do with this reference count mechanism is to emulate the Windows behavior on POSIX systems. Unfortunately, as has been noted, that emulation really isn't very reliable in the face of ill-behaved (i.e. crashing) clients. And I'm pretty sure there isn't a solution to that problem, at least not with the shm_open &etc API. First question: Why not use the shm_open interface on Windows? One possible answer would be that the Windows POSIX support doesn't include the shm_xxx API. And that might even be the answer, since some web searches have led me to suspect that Windows only supports the SysV shared memory API. Which leads to Second question: Why use different implementations on different platforms? Why not use the (admittedly somewhat clumsy to use directly) SysV shared memory API, which is pretty widely supported? It's a little more painful to use, but that's an implementation detail that won't be exposed to library clients. One issue might be kernel limits on min/max size and number of objects; some of the references I have make mention of such but don't provide much detail. On my stock configured SuSE9.3 machine I see max number of identifiers 4096 and max size 33554432 (32768 * 1024). SysV shared memory also doesn't support resizing, which is available when using the shm_open API (I don't know whether that feature is available with the create_file_mapping API). I don't know if that actually matters for this library. Note that one of the capabilities of SysV shared memory is obtaining the current number of attaches. I don't know if it reliably gets decremented when an attached process exits without explicitly detaching though; I would hope so, but haven't verified it. If so, that could provide a correct reference count implementation. Note also that the size is also recorded, which I think would eliminate the need for the header in the current shared_memory implementation. Hm. If the SysV API provides a correct reference count implementation, and is widely supported, including on Windows, then using that API would permit either lifetime behavior to be implemented (possibly even configurable). Of course, we have yet to have the discussion of which behavior is actually preferable. But it is certainly the case that the present situation, where the behavior is not documented and is buggy on non-Windows platforms, is not ideal.

5 13

Dimensional Analysis
by Paul Mensonides 11 Mar '06

11 Mar '06

I was playing around with Dave's and Aleksey's dimensional analysis example, and I decided that I couldn't leave well enough alone. To that end, I decided to reimplement it so that combining units could be written using infix operator notation. Anyway, the attached file implements it so the following works (on EDG). #include "dim.hpp" #include <iostream> int main(void) { using namespace dim; using dim::time; // annoying names from C headers! DIM( length / time ) velocity; DIM( velocity / time ) acceleration; DIM( mass * acceleration ) force; quantity<float, DIM(mass)> m = 5.0f; quantity<float, DIM(acceleration)> a = 9.8f; quantity<float, DIM(force)> f = m * a; std::cout << "force = " << f.value() << '\n'; return 0; } I find the syntax like DIM( mass * length / (time * time) ) force; mildly amusing, so I thought I'd share. Regards, Paul Mensonides

3 7

[optional] swapping initialized with uninitialized
by Vaclav Vesely 11 Mar '06

11 Mar '06

Hi, When two optionals (one initialized and other not) are swaped, then the one value in uninitialized optional is constructed and the other one in initialized optional is destructed. IMHO it's not necessary. It should be more effective to swap optional's m_storages without any values constructiona and destruction. Is there any reason for the current implementation? Regards, Vaclav

3 2

boost::serialization - problem move to 1.33.1 version
by Алексей Обухов 11 Mar '06

11 Mar '06

Hello all ! I write small test program using boost::serialization. If I use boost 1.32 - compilation OK. But I change boost to 1.33.1 - compiler rise error :( Where I be mistaken ? Compiler messages: OtherTest.cpp w:\Devil\Projects\PLTools2_boost_1.33.1\boost\boost\serialization\export.hpp(146) : error C2027: use of undefined type 'boost::serialization::extended_type_info_null' with [ T=BoostTest2::A ] w:\Devil\Projects\PLTools2_boost_1.33.1\boost\boost\serialization\export.hpp(146) : see reference to class template instantiation 'boost::serialization::extended_type_info_null' being compiled with [ T=BoostTest2::A ] w:\Devil\Projects\PLTools2_boost_1.33.1\boost\boost\serialization\export.hpp(145) : while compiling class-template member function 'void boost::archive::detail::guid_initializer::export_register(const char *)' with [ T=BoostTest2::A ] w:\Devil\Projects\PLTools2_boost_1.33.1\Tests\OtherTest\boost_test_2.h(64) : see reference to class template instantiation 'boost::archive::detail::guid_initializer' being compiled with [ T=BoostTest2::A ] Results This error point to line: (full code below) BOOST_CLASS_EXPORT_GUID( BoostTest2::B, "ClassB" ) OtherTest.cpp: #include <iostream> #include <sstream> #include <boost/serialization/vector.hpp> #include <boost/serialization/string.hpp> #include <boost/serialization/nvp.hpp> #include <boost/archive/xml_iarchive.hpp> #include <boost/archive/xml_oarchive.hpp> #include <boost/serialization/export.hpp> #include <boost/serialization/base_object.hpp> #include "boost_test_2.h" using namespace BoostTest2; /// ..... boost_test_2.h #pragma once namespace BoostTest2 { class A { public: A() {m_A = 1;} virtual ~A() { } int m_A; }; class B : public A { public: B() {m_B = 2;} virtual ~B() {} int m_B; }; class C : public A { public: C() {m_C = 3;} virtual ~C() {} int m_C; }; } #define BOOST_NVP(name,value) \ boost::serialization::make_nvp(name, value) namespace boost { namespace serialization { template<class Archive> void serialize(Archive & ar, BoostTest2::A &a, const unsigned int version) { ar & BOOST_NVP( "A", a.m_A ); } template<class Archive> void serialize(Archive & ar, BoostTest2::B &b, const unsigned int version) { ar & boost::serialization::make_nvp( "A", boost::serialization::base_object<A>(b) ); ar & BOOST_NVP( "B", b.m_B ); } template<class Archive> void serialize(Archive & ar, BoostTest2::C &c, const unsigned int version) { ar & boost::serialization::make_nvp( "A", boost::serialization::base_object<A>(c) ); ar & BOOST_NVP( "C", c.m_C ); } } } BOOST_CLASS_EXPORT_GUID( BoostTest2::B, "ClassB" ) BOOST_CLASS_EXPORT_GUID( BoostTest2::C, "ClassC" ) BOOST_IS_ABSTRACT( BoostTest2::A ) BOOST_CLASS_VERSION( BoostTest2::A , 0 )

1 0

Re: [boost] [interprocess] shared memory lifetime
by Olaf van der Spek 10 Mar '06

10 Mar '06

On 3/10/06, Kim Barrett <kab(a)irobot.com> wrote: > >One option is to use a memory mapped file to emulate posix-like behavior > >in windows. But it surely will be slower (the OS will dump data to file) > >and the unlink is not trivial. > > That does seem like it would be bad if true. Are there any file > manipulation options that might be used to control that? I don't know, > but I'll ask around and do some reading and see what I can find. > > [Later] As long as you don't flush the file it would seem like a memory > mapped file should be no different from a shared memory in performance, > the only difference is where data gets written if it needs to be swapped > out. The big downside to using an actual file would be that you need a > place in the file system to put it. Then you get into the whole name of > the shared memory portability problem again. The OS will probably write dirty pages to disk even before paging them out so that if it needs to page them out it doesn't have to also do a disk write.

1 0

[Announce] Mailing list downtime this Saturday
by Doug Gregor 10 Mar '06

10 Mar '06

All Boost mailing lists hosted at lists.boost.org will be down from 8pm to midnight EST this Saturday, March 11th while we migrate to a new, dedicated mail server provided by the Open Systems Lab at Indiana University. During this down time, the mail archives at lists.boost.org will be unavailable and any mail sent to lists.boost.org will be queued and forwarded. We ask that members refrain from posting messages to lists.boost.org during this time "just in case," although we expect that all mail sent during this time will be forwarded properly. If you should have any questions, please contact me directly at dgregor(a)cs.indiana.edu. Doug Gregor Boost Moderator dgregor(a)cs.indiana.edu

1 0

[FileSystem] Incorrect info returned by filesystem::filesystem_error exception
by james＠massey46.fsnet.co.uk 10 Mar '06

10 Mar '06

Hi, There appears to be problems with boost::filesystem::rename with respect to the information returned by boost:filesystem::filesystem_error exception. See the 2 tests below and attached source code examples. Test 1: Precondition: The "Source" file does not exist and the "Destination" file does exist. Test: Perform rename(Source,Destination); Expected Result: The rename fails because the Source file does not exist. Actual Result: A boost filesystem exception is thrown and the output is what text: boost::filesystem::rename: "Source", "Destination": Success error_code: 1 native_error: 0 The exception information suggests the command was successful. The result of a UNIX "mv" with the same scenario is "mv: cannot stat `Source': No such file or directory" Note that a similar output is displayed if the "Source" and "Destination" files exist before the test is run. Test 2: Precondition: The "Source" and "Destination" files exist. The directory DirWithNoWritePermission exists and has permissions 555. Tests: Perform rename(Source,DirWithNoWritePermission/Anything); Perform rename(Source,Destination); Expected Result: The first rename fails because the directory permissions forbid the rename. The second rename fails because the destination file exists. Actual Result: A boost filesystem exception is thrown and the output is what text: boost::filesystem::rename: "Source", "DirWithNoWritePermission/Anything": Permission denied error_code: 3 native_error: 13 and is the same in both cases. It appears that the exception information displayed for the second test has been "recalled" from the first test. This may be related to the problem shown in Test 1. Test programs run under Suse Enterprise Linux 9 with gcc version 3.3.3, boost 1_33_0 and 1_33_1 Regards James Massey

1 0

[statechart] - error with templated states (no type 'inner_contex_type')
by Oliver.Kowalke＠infineon.com 10 Mar '06

10 Mar '06

Hello, I get another error if I use templated states (gcc-3.4.6). 'simple_state.hpp:403: error: no type named inner_context_type in struct proto<int>' Instantiation of state_machine:: initiate() causes the error. with regards, Oliver #include <iostream> #include <stdexcept> #include <boost/statechart/simple_state.hpp> #include <boost/statechart/state_machine.hpp> namespace sc = boost::statechart; template< typename S > struct proto; template< typename S > struct sm : public sc::state_machine< sm< S >, proto< S > > {}; template< typename S > struct inactive; template< typename S > struct proto : public sc::simple_state< proto< S >, sm< S >, inactive< S > > {}; template< typename S > struct inactive : public sc::simple_state< inactive< S >, proto< S > > {}; int main() { try { sm< int > fsm; fsm.initiate(); // <<== causes compiler error return 0; } catch ( std::runtime_error const& e) { std::cerr << e.what() << std::endl; } catch (...) { std::cerr << "unhandled exception" << std::endl; } return -1; }

3 3

[optional] swapping initialized with uninitialized
by Vaclav Vesely 10 Mar '06

10 Mar '06

Hi, When two optionals (one initialized and other not) are swaped, then the one value in uninitialized optional is constructed and the other one in initialized optional is destructed. IMHO it's not necessary. It should be more effective to swap optional's m_storages without any values constructiona and destruction. Is there any reason for the current implementation? Regards, Vaclav

1 0

in_place_factory0
by Vaclav Vesely 10 Mar '06

10 Mar '06

Hi, is there any reason for missing in_place_factory without arguments? I would like to write optional<my_class> mc(in_place()); I've wrote the code for this. But I can't modify the macro hell in "in_place_factory.hpp". Regards, Vaclav class in_place_factory0: public in_place_factory_base { public: in_place_factory0() {} template<class T> void apply(void* address) const { new(address) T(); } }; in_place_factory0 in_place() { return in_place_factory0(); }

1 0