- Boost-users - lists.boost.org

[boost-users][test] Does memory leak detection work in linux?
by Andrey Torba 17 Mar '10

17 Mar '10

Hi all, Does memory leak detection work in linux? -- Regards, Andrey

3 4

How to write a config file parser with boost::spirit that supports line comments?
by Lars Rohwedder 17 Mar '10

17 Mar '10

Hi there, I want to write a parser for config files having many entries with this format: <name> "{" <token> [ <token> ... ] "}" The format is "free-form" so there can be line breaks and other whitespaces within each entry between the tokens etc. <name> is chset_p("a-zA-Z_"), <token> is chset_p("a-zA-Z0-9_.-"). This parser works fine so far. I extended the format to allow even whitespaces etc. in the tokens. So <token> is changed to (I omit the assign_a actions): ( +chset_p("a-zA-Z0-9_.-") | ('"' >> +(!ch_p('"')) >> '"') | ('\'' >> +(!ch_p('\'')) >> '\''; which also works fine as I tested. (I know, newlines would be allowed in these tokens but I hope noone use that.) But now I want to extend the format to also allow line comments, starting with '#', ending with '\r' or '\n', and I failed: Just scanning for '#' also recognized '#' within a quoted token. But that is not the start of a comment. :-( But first searching for quotes also catched unbalanced quote chars in comments and confused the parser. What would be a working solution for this - IMHO still simple - format? (I should mention that I am stuck with boost 1.35 so I cannot use spirit2 or all the other new features in the current release. :-(( ) Lars R.

1 0

custom allocators Re:pool_alloc
by B Hart 17 Mar '10

17 Mar '10

I am not understanding the use of pool with std::set. i.e. the release_memory() function referenced in last example shown on: http://www.boost.org/doc/libs/1_42_0/libs/pool/doc/interfaces.html seems to do nothing. I call GetProcessMemoryInfo() and don't see any change after the set has gone out of scope. maybe someone could relate what happens when set goes out of scope and release_memory() isn't called versus set goes out of scope and release_memory() is called. I want to do something like: 1.std::set<unsigned int, std::less<unsigned int>, boost::pool_allocator<unsigned int>> v; 2. for(int i=0; i<(K); i++) { v.insert(i); } 3. use v 4. release_memory() on v //dump all set elements in one action 5. repeat 1-4 with a new set

1 0

[serialization] xml files that can be hand-edited
by Paul 17 Mar '10

17 Mar '10

Hi, I'd like to build some insurance into my XML files. If there is some sort of problem in the future (either with boost, or my application), I'd like to be able to hand-edit the serialized XML archive files. Even better would be to allow XML-tools to manipulate a batch of the XML files, but one thing at a time. One big problem with the current XML format is that the class_id="3" tracking_level="0" version="0" preamble is only included the FIRST time a class is dumped to the file. Example: <brother class_id="3" tracking_level="0" version="2"> .. etc, and then later in the file ... <sister> ... etc ... If something happened (eg impossible-to-upgrade-format-change) and I needed to manually move the <sister> item above the <brother> item (maybe into a different branch of tags), then I would need to KNOW that the brother class is the same as the sister class (which is not obvious) AND move the preamble into the sister tag. I would also need to re-number all the class_ids, but those appear to be only used if I'm tracking a pointer AND I have multiple instances of that pointer in the file. If the item is not a pointer, it looks like the class_id is *always* thrown away. So, I'm thinking of: a) never allow tracking (its a compromise to give me the next item) b) never writing out class_id or tracking_level c) always writing out the version the result would look like this: <brother version="2"> .. etc, and then later in the file ... <sister version="2"> That way items in the XML file could be rearranged by hand and I would not need to do the difficult task of adjusting the tag attributes. Does this sound sensible? I had a look into inheriting and creating my own archive, but it appears all of the changes required happen in the basic_oarchive_impl, which seems pretty fundamental and hard to override. Will I need to create and maintain my own patch in order to "hack" this change. Thanks, Paul

2 1

[date time] convert UTC date-time to run-time-determined local time zone
by eburkitt＠gmail.com 17 Mar '10

17 Mar '10

I'm developing a program that will perform tasks on behalf of clients in various (US) time zones. It receives a UTC date-time from the client, along with an offset from UTC and a flag as to whether the client's area observes DST. The program needs to convert the time into both its own local time and the client's local time. I have not been able to come up with a reasonable way to do either in the date-time library. The only way I can find in the library to convert UTC to local time requires me to know the time zone at compile time. For the client I have the offset from UTC and its local DST observance policy. (I can stipulate at this point that the client uses US DST rules.) For the program itself, I have only what I can query from the computer on which it runs. Am I missing something obvious?

1 0

Re: [Boost-users] How to forward to a list of boost::function?
by OvermindDL1 17 Mar '10

17 Mar '10

-----Original Message----- From: John Ky <newhoggy(a)gmail.com> Sent: Tuesday, March 16, 2010 5:29 AM To: boost-users <boost-users(a)lists.boost.org> Subject: Re: [Boost-users] How to forward to a list of boost::function? > > Hi OvermindDL1, > > I'm having trouble finding any information on this (or some of the other alternatives presented here). For example, the following page: > > http://www.boost.org/doc/libs/1_42_0/libs/fusion/doc/html/fusion/functional… > > links to a broken link: > > http://www.boost.org/doc/libs/1_42_0/libs/functional/forward/doc/html/index… > > Anywhere I can get to the documentation on this? The work was started in 1.42, however it is finished in the trunk. Look at the functional docs on the trunk and they exist there and all code there works well. Sorry that I cannot supply actual links, not at computer (am on my phone currently).

2 1

How to pass parameters to manage_shared_memory.construct() in Boost.Interprocess
by Cuneyt Taskiran 16 Mar '10

16 Mar '10

I've stared at the Boost.Interprocess documentation for hours but still haven't been able to figure this out. In the doc, they have of creating a vector in shared memory like so: //Define an STL compatible allocator of ints that allocates from the managed_shared_memory. //This allocator will allow placing containers in the segment typedef allocator<int, managed_shared_memory::segment_manager> ShmemAllocator; //Alias a vector that uses the previous STL-like allocator so that allocates //its values from the segment typedef vector<int, ShmemAllocator> MyVector; int main(int argc, char *argv[]) { //Create a new segment with given name and size managed_shared_memory segment(create_only, "MySharedMemory", 65536); //Initialize shared memory STL-compatible allocator const ShmemAllocator alloc_inst (segment.get_segment_manager()); //Construct a vector named "MyVector" in shared memory with argument alloc_inst MyVector *myvector = segment.construct<MyVector>("MyVector")(alloc_inst); Now, I understand this. What I'm stuck is how to pass a second parameter to `segment.construct()` to specify the number of elements. The interprocess document gives the prototype for `construct()` as MyType *ptr = managed_memory_segment.construct<MyType>("Name") (par1, par2...); but when I try MyVector *myvector = segment.construct<MyVector>("MyVector")(100, alloc_inst); I get compilation errors. My questions are: 1. Who actually gets passed the parameters `par1, par2` from `segment.construct`, the constructor of the object, e.g. `vector`? My understanding is that the template allocator parameter is being passed. Is that correct? 2. How can I add another parameter, in addition to `alloc_inst` that is required by the constructor of the object being created in shared memory? Thanks!

4 4

[serialization] polymorphic portable binary example not working.
by Mostafa 16 Mar '10

16 Mar '10

Hi all, Can someone please shed some light on this. I'm trying to compile the polymorphic portable binary serializor in the examples directory and getting compiler errors. Some I managed to fix because a file seemed to have been deleted and its class moved to another file, but this one has stumped me. In polymorphic_portable_binary_oarchive.hpp, its complaining about: #include <boost/archive/detail/polymorphic_oarchive_dispatch.hpp> which is needed for: typedef boost::archive::detail::polymorphic_oarchive_dispatch< portable_binary_oarchive > polymorphic_portable_binary_oarchive; Obviously polymorphic_oarchive_dispatch.hpp is missing and there isn't a class definition for polymorphic_oarchive_dispatch in any other files. (I've downloaded the boost source code so I'm not relying on any packages.) Any help on how to resolve this would be appreciated. Thanks, Mostafa

2 13

scatter error
by alev mutlu 16 Mar '10

16 Mar '10

Hi The code below compiles fine , but gives the [ceng1:23288] *** Process received signal *** [ceng1:23288] Signal: Segmentation fault (11) [ceng1:23288] Signal code: Address not mapped (1) [ceng1:23288] Failing at address: 0xfffffffffffffff9 [ceng1:23288] [ 0] /lib64/libpthread.so.0 [0x3baf40de80] [ceng1:23288] [ 1] /usr/lib64/libstdc++.so.6(_ZN9__gnu_cxx18__exchange_and_addEPVii+0x2) [0x38d9cb7672] [ceng1:23288] [ 2] /usr/lib64/libstdc++.so.6(_ZNSs6assignERKSs+0x9b) [0x38d9c9ca4b] [ceng1:23288] [ 3] test-boost(_ZN5fieldaSERKS_+0x47) [0x4219a1] [ceng1:23288] [ 4] test-boost(_ZNSt6__copyILb0ESt26random_access_iterator_tagE4copyIPK5fieldPS3_EET0_T_S8_S7_+0x3c) [0x4221e0] [ceng1:23288] [ 5] test-boost(_ZSt10__copy_auxIPK5fieldPS0_ET0_T_S5_S4_+0x29) [0x422225] [ceng1:23288] [ 6] test-boost(_ZNSt13__copy_normalILb0ELb0EE6copy_nIPK5fieldPS2_EET0_T_S7_S6_+0x25) [0x42224d] [ceng1:23288] [ 7] test-boost(_ZSt4copyIPK5fieldPS0_ET0_T_S5_S4_+0x2d) [0x42227d] [ceng1:23288] [ 8] test-boost(_ZN5boost3mpi6detail12scatter_implI5fieldEEvRKNS0_12communicatorEPKT_PS7_iiN4mpl_5bool_ILb0EEE+0x90) [0x42a652] [ceng1:23288] [ 9] test-boost(_ZN5boost3mpi7scatterI5fieldEEvRKNS0_12communicatorEPKT_PS6_ii+0x50) [0x42b184] [ceng1:23288] [10] test-boost(_ZN5boost3mpi7scatterI5fieldEEvRKNS0_12communicatorERKSt6vectorIT_SaIS7_EEPS7_ii+0x55) [0x42b1fd] [ceng1:23288] [11] test-boost(main+0x231) [0x4208a3] [ceng1:23288] [12] /lib64/libc.so.6(__libc_start_main+0xf4) [0x3bae81d8b4] [ceng1:23288] [13] test-boost(__gxx_personality_v0+0x169) [0x420489] [ceng1:23288] *** End of error message *** mpirun noticed that job rank 0 with PID 23288 on node ceng1-ib exited on signal 11 (Segmentation fault). run time errors. Is this due to miss-use of boost functions or am I doing wrong with memory allocation? #include <boost/mpi/environment.hpp> #include <boost/mpi/communicator.hpp> #include <boost/mpi/collectives.hpp> #include <boost/thread/barrier.hpp> #include <boost/thread/mutex.hpp> #include <boost/config.hpp> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/access.hpp> #include <boost/serialization/string.hpp> #include <string> #include <iostream> #include <vector> #include <fstream> #define MASTER 0 int arraySize = 7; using namespace std; namespace mpi = boost::mpi; struct field { string field_name; string field_type; string field_ref; string field_colType; }; namespace boost { namespace serialization{ template<class Archive> void serialize(Archive & ar, struct field & f, unsigned int version){ ar & f.field_name; ar & f.field_type; ar & f.field_ref; ar & f.field_colType; } } } int main(int argc, char* argv[]) { mpi::environment env(argc, argv); mpi::communicator world; int m_mySize, w_mySize, kk,i; struct field *relAllValues; struct field *relMyValues; vector<struct field> C; struct field c; w_mySize = arraySize / world.size(); cout << "w_mySize = " << w_mySize << endl; if (world.rank() == MASTER){ for(int i = 0; i < arraySize; i++){ c.field_name = "field_name=" + i; c.field_type = "field_type=" + i; c.field_ref = "field_ref=" + i; c.field_colType = "field_colType=" + i; C.push_back(c); } m_mySize = arraySize / world.size() + arraySize % world.size(); relMyValues = (struct field *) malloc(sizeof(struct field) * m_mySize); relAllValues = (struct field *) malloc(sizeof(struct field) * arraySize); } else{ relMyValues = (struct field *) malloc(sizeof(struct field) * w_mySize); } if (world.rank() == MASTER){ scatter(world, C, relMyValues, w_mySize, 0); } else{ scatter(world, C, relMyValues, w_mySize, 0); } }

2 2

regex - build problem
by Christian Meesters 16 Mar '10

16 Mar '10

Hi, I recently installed boost v1.42, everything seemed fine, but now I'm trying to use the regex-library -- and cannot compile/link my code anymore. gcc tells me that upon linking it cannot find 'libicudata.so.38', 'libicui18n.so.38', and 'libicuuc.so.38' and complaints about a number of undefined references. Indeed I have icu40 not 38 installed on my system. Just providing symbolic links with the appropriate names does not help. I then tried updating my boost build with bjam. But bjam also complaints (with ./bjam --with-regex toolset=gcc -sHAVE_ICU=1 -sICU_PATH=/usr/lib): warning! ICU configuration failed Couldn't find utypes.h in /usr/lib/include/unicode warning: Building Boost.Regex with the optional Unicode/ICU support disabled. There are no further warnings, yet linking still results in undefined references. 'utypes.h' is nowhere on my system and apparently does not come with the icu-dev package. I am running the current Ubuntu (9.10), all updates installed. boost ist compiled on my system, no package involved. And I'm using gcc 4.4 - if that matters. Now, could this be a versions issue? Any hint on where to look further? Any ideas? TIA Christian

2 2