June 2013 - Boost-users - lists.preview.boost.org

Re: [Boost-users] Mutex synchronised wrapper class for boost::interprocess shared memory vector of strings
by Riskybiz 24 Jun '13

24 Jun '13

Date: Wed, 19 Jun 2013 22:05:32 +0200 From: Ion Gazta?aga <igaztanaga(a)gmail.com> To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] Mutex synchronised wrapper class for boost::interprocess shared memory vector of strings Message-ID: <51C20F0C.8090408(a)gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed El 18/06/2013 16:28, Riskybiz escribi?: > I'm trying to write a wrapper class in C++ to give mutex synchronised > interprocess access to a shared memory vector of strings. This is my first > foray into boost::interprocess and allocators are proving something of a > mystery. > > Is anyone able to see why on compilation I get the error: When you build any container in shared memory it must be constructed with an allocator. Change your push function to (not charallocator is passed to MyShmString): void push(const std::string & in) { CharAllocator charallocator(m_segment->get_segment_manager()); scoped_lock<named_mutex> lock(*m_mutex); MyShmString inStr(in.c_str(), charallocator); sharedSegmentVector->push_back(inStr); } Best, Ion Thanks for the reply Ion. I'm now having trouble reading from the vector in the shared memory. I get an exception when calling read(0) or read(1). What I'm trying to do is read "testing testing 123" from the first element of the sharedSegmentVector. Is it necessary to use some particular type conversion technique in the reader function to access a sharedSegmentVector string element? Any help to get this working is much appreciated; I've been trying for more than a month to get to this point! Code below..... Thanks Riskybiz #ifndef SHARED_MEMORY_WRAPPER_H//if not defined already #define SHARED_MEMORY_WRAPPER_H//then define it #include <boost/interprocess/managed_shared_memory.hpp> #include <boost/interprocess/containers/vector.hpp> #include <boost/interprocess/containers/string.hpp> #include <boost/interprocess/allocators/allocator.hpp> #include <boost/interprocess/sync/named_mutex.hpp> using namespace boost::interprocess; class SharedMemoryWrapper { public: SharedMemoryWrapper(const std::string &name, bool server) : m_name(name), m_server(server) { if (server) { named_mutex::remove("named_mutex"); shared_memory_object::remove(m_name.c_str()); //Create shared memory managed_shared_memory m_segment(create_only, m_name.c_str(), 10000); //Create allocators CharAllocator charallocator(m_segment.get_segment_manager()); StringAllocator stringallocator(m_segment.get_segment_manager()); //This string is in only in this process (the pointer pointing to the //buffer that will hold the text is not in shared memory). //But the buffer that will hold "this is my text" is allocated from //shared memory MyShmString mystring(charallocator); mystring = "testing testing 1 2 3"; //This vector is fully constructed in shared memory. All pointers //buffers are constructed in the same shared memory segment //This vector can be safely accessed from other processes. MyShmStringVector *sharedSegmentVector = m_segment.construct<MyShmStringVector>("sharedSegmentVector")(stringallocato r); sharedSegmentVector->push_back(mystring); } else { //Open the shared segment managed_shared_memory m_segment(open_only, name.c_str()); //Find the vector using the c-string name and open it MyShmStringVector *sharedSegmentVector = m_segment.find<MyShmStringVector>("sharedSegmentVector").first; } m_mutex = new named_mutex(open_or_create, "named_mutex"); } ~SharedMemoryWrapper() { if (m_server) { named_mutex::remove("named_mutex"); //Destroy the vector from the managed_shared_memory m_segment->destroy_ptr(sharedSegmentVector);//This will free all strings that the vector contains m_segment->destroy<MyShmStringVector>("sharedSegmentVector");//Destroy the vector //Is it necessary to destroy the allocators here? //Remove the managed_shared_memory; this may fail if the memory does not exist or is mapped or opened by another process shared_memory_object::remove(m_name.c_str()); } delete m_mutex; delete m_segment; } //Writer void write(const std::string &in) { CharAllocator charallocator(m_segment->get_segment_manager()); scoped_lock<named_mutex> lock(*m_mutex); MyShmString shmStr(in.c_str(), charallocator); sharedSegmentVector->push_back(shmStr); } //Reader std::string read(long index) { CharAllocator charallocator(m_segment->get_segment_manager()); scoped_lock<named_mutex> lock(*m_mutex); MyShmString shmStr(sharedSegmentVector->at(index), charallocator); std::string stdStr = std::string(shmStr.begin(), shmStr.end()); return stdStr; } private: //Typedefs typedef boost::interprocess::allocator<char, managed_shared_memory::segment_manager> CharAllocator; typedef boost::interprocess::basic_string<char, std::char_traits<char>, CharAllocator> MyShmString; typedef boost::interprocess::allocator<MyShmString, managed_shared_memory::segment_manager> StringAllocator; typedef boost::interprocess::vector<MyShmString, StringAllocator> MyShmStringVector; bool m_server; std::string m_name; managed_shared_memory *m_segment; MyShmStringVector *sharedSegmentVector; named_mutex *m_mutex; };//class #endif//header guard

2 1

[interprocess]shared_memory_object stores content in disk?
by Adrian Yu 24 Jun '13

24 Jun '13

I'm training multiple networks based on a single database. So to accelerate speed and reduce disk reading, I use shared_memory_object class provided by boost. Since the lab workstation is currently unavailable, I migrated my code to my personal computer. On the lab workstation, the host program successfully reads all data to memory. But on my PC, strangely it creates a file on system drive rather than storing the data in memory. The whole database is about 3.7 GB; the lab workstation has 32 GB memory and runs Windows Server 2008 R2; my PC has 8 GB memory and runs Windows 7. There should be enough memory to store the data. So why? Are there certain ways to force the program to keep all data in memory?

2 1

[Interprocess] Allocation algorithms
by Oodini 24 Jun '13

24 Jun '13

Hello, I need to allocate and fill about 10 big memories areas (several Gb) at the startup of my program. The data will never be modified or freed after during the program lifetime, and will be *only read* an other processes. What allocation algorithm should I take from the ones provided by the library ? Should I create a new one, more efficient (no reallocations, no race conditions...) ? Thanks.

3 3

[Interprocess] Minimal size of a managed_shared_memory
by Oodini 24 Jun '13

24 Jun '13

Hello, A managed_shared_memory is a typedef for a basic_managed_shared_memory using rbtree_best_fit as allocation algorithm. In the documentation : http://www.boost.org/doc/libs/1_53_0/doc/html/interprocess/memory_algorithm… It is said at the and of the page : "In most 32 systems with 8 byte alignment the minimum size of a block is 24 byte". I am on Windows XP SP3 32 bits. The alignement is 8 byte. But if I create a managed_shared_memory with a size lesser than 120 bytes, my app crashes. This value comes from several get_min_size function, the deepest being in the rbtree_best_fit class. So I wonder which value we can grant to the documentation above. How are the 120 bytes used ?

2 1

Boost::Interprocess Documentation request
by Riskybiz 24 Jun '13

24 Jun '13

In the Boost::Interprocess documentation and under the heading "Where is this being allocated?"; is an example of how to create a vector of strings in a managed_shared_memory segment. Link to Boost::Interprocess documentation here: http://www.boost.org/doc/libs/1_53_0/doc/html/interprocess/allocators_contai ners.html#interprocess.allocators_containers.containers_explained Please could this example be extended to demonstrate; 1. How to add string elements to the shared memory vector from a different process? 2. How to access and copy strings from the shared memory vector from a different process? This really would be invaluable to people trying to use Boost::Interprocess for the first time! My particular project has ground to a halt for over a month simply trying to get a working implementation of a Boost::Interprocess mutex synchronised shared memory vector of strings. I would like one process to add strings to the shared memory vector and several other processes to read string data from the vector. Right now I don't know if I'll ever get it to work! Running out of ideas here! Hope someone can help. Thanks, Riskybiz.

1 0

[property_tree] Serializing ptree to/from JSON
by Michael Powell 24 Jun '13

24 Jun '13

Hello, I am writing some code to serialize a package to/from JSON. Overall, it is working well, except for what appears to be a weak link in the ptree chain of events. I take a POCO (for all intents) and populate a ptree, then parse that to JSON using the property_tree JSON parser. No problem. Then I take that same string and want to parse it into another ptree. In and of itself, seems like it is no problem. The new ptree has a count, so I gather the parse worked. However, when I try to use ptree.count(path), or ptree.find(path), neither of these work. They are returning 0 or not_found(), respectively. I may be able to bypass that check for the moment and run it through ptree.get<Type>(path) for the time being, but that seems less desirable. Any insights on ptree along these lines? Thanks! Regards, Michael Powell

3 3

[FunctionTypes] Getting functor/lambda parameters (VS2012)
by Mateusz Loskot 23 Jun '13

23 Jun '13

Hi, I'm using C++ compiler from the latest VS2012. I'm trying this simple technique to determined type of functor or lambda call parameter, it works well: #include <boost/function_types/parameter_types.hpp> int main() { auto foo = [](int) -> void {}; typedef decltype(foo) foo_type; typedef boost::function_types::parameter_types < decltype(&foo_type::operator()) >::type parameter_types; } Next, I move the typedef into class template, in constructor body, it works well too: template <typename T> struct A { A(T t) { typedef typename boost::function_types::parameter_types < decltype(&T::operator()) >::type parameter_types; } }; Finally, I move further with the technique, so I can access parameter types within class template scope: template <typename T> struct B { typedef typename boost::function_types::parameter_types < decltype(&T::operator()) >::type parameter_types; }; This example does compile with clang 3.4 (svn trunk) and GCC 4.8.1, but it does not compile with Visual C++ 11.0. Does anyone understand what C++11 feaure VS2012 is lacking so it prevents compilation of the sample with B template above? Here is complete .cpp file https://gist.github.com/mloskot/5846468 Best regards, -- Mateusz Loskot, http://mateusz.loskot.net

1 0

install with b2 - conflicting documentation
by Kirk Joppy 23 Jun '13

23 Jun '13

Hello, I am trying to build the regex library which I will then use through MinGW. I already installed MinGW and added ..\MinGW\bin to my system PATH. I now want to build regex and put it somewhere. I am aware of two different pieces of documentation detailing how to do this: http://www.boost.org/doc/libs/1_53_0/more/getting_started/windows.html#prep… http://www.boost.org/doc/libs/1_53_0/doc/html/bbv2/installation.html The main difference seems to be whether you should be in BOOST_ROOT or BOOST_ROOT\tools\build\v2. Which is it? In either case I gather than I should do this $ bootstrap.bat gcc and in fact I've done that in both BOOST_ROOT and BOOST_ROOT\tools\build\v2 with apparent success. I just don't know what I'm supposed to do next. I tried running $ b2 --prefix=H:\boost-test --with-regex install from within BOOST_ROOT\tools\build\v2, and that put some stuff into H:\boost-text, but I couldn't find any actual library files. Before I screw around any more I thought I'd ask knowledgeable folks for the right way to do this. Thank you very much, kjoppy

2 2

An example: how I learned to iterate through property trees (ptree)
by Cook, Rich 22 Jun '13

22 Jun '13

I just wanted to post a program I wrote that creates a boost::ptree, saves it to JSON, reads it back, then iterates through the tree and prints its nodes. #include <boost/property_tree/ptree.hpp> #include <boost/property_tree/json_parser.hpp> #include <string> #include <iostream> using namespace std; using boost::property_tree::ptree; string indent(int level) { string s; for (int i=0; i<level; i++) s += " "; return s; } void printTree (ptree &pt, int level) { if (pt.empty()) { cerr << "\""<< pt.data()<< "\""; } else { if (level) cerr << endl; cerr << indent(level) << "{" << endl; for (ptree::iterator pos = pt.begin(); pos != pt.end();) { cerr << indent(level+1) << "\"" << pos->first << "\": "; printTree(pos->second, level + 1); ++pos; if (pos != pt.end()) { cerr << ","; } cerr << endl; } cerr << indent(level) << " }"; } return; } int main(int, char*[]) { // first, make a json file: string tagfile = "testing2.pt"; ptree pt1; pt1.put("object1.type","ASCII"); pt1.put("object2.type","INT64"); pt1.put("object3.type","DOUBLE"); pt1.put("object1.value","one"); pt1.put("object2.value","2"); pt1.put("object3.value","3.0"); pt1.put("object4.type", "ASCII"); pt1.put("object4.value", "shallow"); pt1.put("object4.subobject1.type", "ASCII"); pt1.put("object4.subobject1.value", "deep child"); pt1.put("object4.subobject1.subsubobject.type", "ASCII"); pt1.put("object4.subobject1.subsubobject.value", "deep child"); pt1.put("object5.type", "ASCII"); pt1.put("object5.value", "shallow"); write_json(tagfile, pt1); ptree pt; bool success = true; //try { read_json(tagfile, pt); printTree(pt, 0); cerr << endl; return success; } PROGRAM OUTPUT: rcook@rzbeast (blockbuster): ./ptree { "object1": { "type": "ASCII", "value": "one" }, "object2": { "type": "INT64", "value": "2" }, "object3": { "type": "DOUBLE", "value": "3.0" }, "object4": { "type": "ASCII", "value": "shallow", "subobject1": { "type": "ASCII", "value": "deep child", "subsubobject": { "type": "ASCII", "value": "deep child" } } }, "object5": { "type": "ASCII", "value": "shallow" } } -- ✐Richard Cook ✇ Lawrence Livermore National Laboratory Bldg-453 Rm-4024, Mail Stop L-557 7000 East Avenue, Livermore, CA, 94550, USA ☎ (office) (925) 423-9605 ☎ (fax) (925) 423-6961 --- Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept. (opinions expressed herein are mine and not those of LLNL)

2 1

[smart_ptr] enable_shared_from_this and shared_ptr to a base class
by Adam Romanek 21 Jun '13

21 Jun '13

Hi, I have a question related to using shared_from_this() and a shared_ptr to a base class. Here's a modified version of the enable_shared_from_this usage example from its documentation web page: #include <boost/enable_shared_from_this.hpp> #include <boost/shared_ptr.hpp> #include <cassert> class X { public: virtual void foo() = 0; }; class Y: public X, public boost::enable_shared_from_this<Y> { virtual void foo() { shared_from_this(); } }; int main() { X* x_ptr = new Y; boost::shared_ptr<X> sp(x_ptr); sp->foo(); } First, class Y is now derived from class X. Secondly, shared_ptr's template argument is now X and the parameter to the shared_ptr's constructor is now a pointer to X. Now, calling shared_from_this() in X and/or Y throws boost::bad_weak_ptr. Is this correct behavior? Is it documented somewhere? I know that the example is a bit unrealistic since most of the time shared_ptr is constructed directly from the pointer returned by 'new Y' or created by make_shared() but still, this is something that bothers me. WBR, Adam Romanek

1 0