- Boost - lists.boost.org

pool allocation
by Stefan Strasser 24 Sep '09

24 Sep '09

I took a closer look at Boost.Pool today to replace some default allocation of small objects. what I found there was a little disappointing, at least for my use case. the object_pool has 2 major downsides: - no dynamic block allocation - destroying one(!) object is O(N), with N being the number of unused objects in the pool so I wrote some code for my own use. I'm not suggesting this code should be added to boost, but something like it. struct A{ ... }; { pool<A,100> pool; A *a; for(int c=0;c<250;++c){ a=pool.new_(); // O(1) } pool.delete_(a); //O(1), destructs } //destruction of 249 objects, O(N) this allocates 3 memory blocks. but it also has one pointer memory overhead per object, to achieve constant time deletion. but you can use that pointer while the object is in use, which was what lead to this code: I needed to store unused boost.intrusive nodes: struct A : intrusive::[s]list_base_hook<>{ ... }; { [s]list_pool<A,[s]list_base_hook<>,100> pool; //same as above } this has no memory overhead, as the list hook can be used by the pool. here's the code: http://www.boostpro.com/vault/index.php?action=downloadfile&filename=intrus…

3 4

[asio] Mixing of sync and async calls
by Roland Schwarz 24 Sep '09

24 Sep '09

Is it legal to mix async and sync function calls? (I was not able to find the answer in the doc.) E.g. socket.async_read_some(buf); socket.write_some(otherbuf); Also when in a blocking sync call, will it be possible to "cancel" it from another thread, e.g. by posting to the io_service which will invoke a close on the socket? Thank you in advance. -- _________________________________________ _ _ | Roland Schwarz aka. speedsnail |_)(_ | sip:speedsnail@ekiga.net | \__) | mailto:roland.schwarz@chello.at ________| http://www.blackspace.at

2 2

[BGL] read_graphviz()/read_graphml() and subgraphs
by Giuseppe Di Guglielmo 24 Sep '09

24 Sep '09

Dear all, I read on the 'read_graphviz()' page in BGL library doc http://www.boost.org/doc/libs/1_40_0/libs/graph/doc/read_graphviz.html * read_graphviz treats subgraphs as syntactic sugar. It does not reflect subgraphs as actual entities in the BGL. Rather, they are used to shorten some edge definitions as well as to give a subset of all nodes or edges certain properties. For example, the DOT graphs digraph { a -> subgraph {b -> c} -> e } and digraph { a -> b -> e ; a -> c -> e ; b -> c} are equivalent. Is this actually true? Can I read sub-graph from DOT description and obtain a subgraph object? http://www.boost.org/doc/libs/1_40_0/libs/graph/doc/subgraph.html Eventually are subgraph supported by 'read_graphml()' function? http://www.boost.org/doc/libs/1_40_0/libs/graph/doc/read_graphml.html Thanks, G.

1 0

[multi_index] Missing mutex lock?
by Anatoly Pinchuk 23 Sep '09

23 Sep '09

Possible traversal and modifications of a safe container iterators list inside safe_mode::detach_equivalent_iterators function are not protected with locking like the following one in safe_iterator_base::detach function. #if defined(BOOST_HAS_THREADS) boost::detail::lightweight_mutex::scoped_lock lock(cont->mutex); #endif Unless some kind of synchronization is provided by callers of the safe_mode API, the list above can be traversed/modified and modified concurrently in BOOST_HAS_THREADS environment. It would be great if somebody can confirm or deny that there is the access synchronization problem. Regards, Anatoly

3 5

Re: [boost] scoped_ptr deleter
by Berserker 22 Sep '09

22 Sep '09

Stewart, Robert wrote: > That change creates multiple, distinct classes. > shared_ptr<T,boost::checked_deleter<T>> is not the same as shared_ptr<T,my_deleter<T>>. > By embedding the deleter in the shared_ptr with type erasure, > shared_ptr<T> is shared_ptr<T> is shared_ptr<T>, regardless of the deleter. Sure and that's the point :) As I wrote, since the "noncopyable nature" of scoped_ptr (shared_ptr is copiable by design), the creation of distinct classes isn't a problem. I think that the "real problem" here is to provide a solution that doesn't require extra memory overhead. _________________________________________________________________ Gossip, Sport, Notizie... Accendi la Messenger TV! http://www.messenger.it/messengerTV.aspx

9 24

Re: [boost] [Boost-users] Parallel BGL status in Boost
by Craig Henderson 22 Sep '09

22 Sep '09

> Nick Edmonds wrote: > >> On Sat, Sep 12, 2009 at 9:42 PM, Nick Edmonds > <ngedmond(a)cs.indiana.edu> wrote: > >> > >>> yet). There's also a very nascent project to do multi-core/highly > >>> multi-threaded BGL, starting from the BGL sources as opposed to > PBGL. > >> > >> Any pointers ? It would be nice if they look at using the map-reduce > >> library (boost candidate) > >> > > It's very early stages so no, no docs/code yet, it's basically some > new PhD students in my lab with me and another post-doc mentoring > them. I would anticipate that map-reduce is too > coarse-grained/BSP-like to capture the fine grained dependencies we > need to capture to get good performance in a memory-bandwidth bound > domain like sparse graphs, but I'll mention it and see if they want to > write a sample algorithm with it and run some numbers. We're > investigating a few different options for organizing/parallelizing the > work, including starting from scratch. > > It would be nice to see how the map-reduce library performs though. > Do you have any performance data for using it on highly multi-threaded > architectures (Sun Niagara, Cray XMT, anything with >8 cores really)? > Hi Nick, the Map/Reduce library is available in the sandbox and in the vault. The performance is as much down to the implementation of the map and reduce tasks as the framework itself. Jose has compared performance of my library with Standford University's Pheonix implementation and the performance is comparable (http://lists.boost.org/Archives/boost/2009/08/155759.php) I have tested the library on MS Windows on servers with 16-cores and performance has scaled well - if you get the input's segregated sufficiently to avoid contention. I don't yet have any benchmarks, but am looking for some suitable algorithms if you're interested in contributing. Regards -- Craig

1 0

Creating a new toolset for System i *SRVPGMs
by Jon Brisbin 22 Sep '09

22 Sep '09

We're integrating our C++ development with our RPG/400 development and since we use Boost on the PC end, we'd really like to use Boost on the AS400/System i integration stuff. I can successfully compile service programs on the 400 that use boost header-only libraries (with some minor tweaking of the Boost source...I'm finding various things that have to be changed to accommodate the 400 compiler) using the "ixlc" compiler. I think what I need to do to use Boost.Build on the 400 is to create a new toolset that is like gcc, but uses the "ixlc" and "ixlclink" commands to compile *MODULEs and *SRVPGMs. I'm just not familiar enough with Boost.Build to figure out how to do this. I'm actually a Java programmer and not even very experienced in C++, though I've been a unix geek for a long time, so I'm very familiar with the build tools. Since the library/file(member) convention of working with these files on the 400 doesn't translate directly across to files on a unix-like filesystem, I'm thinking I'll have to have directories that correspond to libraries and the source file basename that corresponds to the object name. For example, source file: cpplib/mycpptest.cpp would be compiled using "ixlc" like: ixlc -ocpplib/mycpptest -qdbgview=all -qprint (other options) cpplib/ mycpptest.cpp this would create a *MODULE object called "MYCPPTEST" in library "CPPLIB". How would I translate this to a boost build toolset that I could use by writing a Jamfile like the following...? module cpplib/mycpptest : cpplib/mycpptest.cpp ; srvpgm cpplib/mycpptest : cpplib/mycpptest cpplib/myothrmod ; at some point, I'd probably like to specify a default library, but I can probably do that by using variables in my project's root file. Can anyone help me with this, even though you may not have direct experience with the AS400 (it would be great if there were boost developers who worked on or with 400's, though! ;)? Jon Brisbin Portal Webmaster NPC International, Inc.

1 0

[serialization] serialize shared_from_this() pointer
by Zoran Cvetkov 22 Sep '09

22 Sep '09

Does someone have experience serializing a class that inherits from a shared_from_this() pointer? If I just do nothing I am getting 0 pointer during deserialization. What would be the proper syntax for serializing it, if it's needed?

1 0

review request: addition to type_traits library of is_less_comparable<T> and others
by Frédéric Bron 22 Sep '09

22 Sep '09

Hi folks, I would like to propose to your review the following addition to the type_traits library. The addition is attached to this email and also available in the sandbox at https://svn.boost.org/trac/boost/browser/sandbox/type_traits The purpose of the addition is to add type traits to detect if type T is comparable in the sens of <, <=, >, >=, == or != operators, i.e. if t1<t2 has a sens when t1 and t2 are of type T (same for <=, >, >=, ==, !=). The following traits are added: is_equal_to_comparable<T> is_greater_comparable<T> is_greater_equal_comparable<T> is_less_comparable<T> is_less_equal_comparable<T> is_not_equal_to_comparable<T> The names are based on the corresponding names of the standard template library (<functional> header, section 20.3.3 of the standard). The compilation has been tested with g++ 4.3.2 and icpc 10.0 on linux 64 bits, g++ 3.4.4 on cygwin, mingw32 3.4.4 on cygwin. I must add that I did not invent the code (I obtained some help from the boost mailing list) but I put it together in a form that should be acceptable to boost. Regards, Frédéric

8 23

Possible submission: Dynamic library policy loading library?
by Rob Meijer 22 Sep '09

22 Sep '09

In the Open Computer Forensics Architecture library ( http://sourceforge.net/projects/ocfa/ ), a simple template sub library is used for policy loading with shared libraries. This stuff is currently buried in a large library, and it would seem prety usefull to extract this sub library into an enviroment (like boost) where it can be used within a wider scope. To show apptoximately how this sub library is used, I've taken a few lines from a part of OcfaLib that uses this functionality: The below example is used to load an implementation of the abstract TreeGraphFactory class within a loadable library "libtesttreegraph.so" with a constructor function named "constructor" in it. std::string module="testtreegraph"; //Define a type for the constructor. typedef SinglePointerConstructor< TreeGraphFactory, std::map< std::string,misc::Scalar > > CTreegraphType; //Define a type for the policy loader. typedef PolicyLoader<CTreegraphType> PLTreeGraph; //Create a new policy loader object for a given module and constructor function name. PLTreeGraph *mPolicyLoader=new PolicyLoader<CTreegraphType>(module,"constructor"); //Fetch the constructor from the policy loader. CTreegraphType *constructor = mPolicyLoader->constructor(); //Construct an object using the constructor TreeGraphFactory * mActiveFileSystem =(*constructor)(attributes); For more info, the relevant headers in the ocfa source tree are: OcfaLib/inc/misc/PolicyLoader.hpp OcfaLib/inc/misc/VoidToFunction.hpp It probably needs some work on portability and needs more constuctor templates to be fully generically usable, but I feel this has the potential to be a very usefull extention to boost. Do you people feel this should be further explored and submitted as a possible extension for boost? Rob

1 0