April 2010 - Boost - lists.preview.boost.org

[shared_ptr] Bug in perfect forwarding test?
by John Maddock 26 Apr '10

26 Apr '10

Now that rvalue refs are enabled for VC++10 the perfect forwarding test fails: http://tinyurl.com/367o3ab This looks like a legitimate failure to me, as the constructor to Y requires an "int&" non-const reference, but the make_shared_ptr function passes a "const int&". Ah... actually this may be to do with the fact that the header is predicated on BOOST_HAS_VARIADIC_TMPL as well as BOOST_HAS_RVALUE_REFS where as the test case is only predicated on BOOST_HAS_RVALUE_REFS? BTW there's an rvalue-refs related failure with gcc-4.5 in C++0x mode as well: http://tinyurl.com/37quwbj Cheers, John.

2 1

[1.43] Wrong version in version.hpp
by John Maddock 26 Apr '10

26 Apr '10

Just a note to point out that the release branch still has the version number set to 1.42 in boost/version.hpp. John.

2 1

[async] messages threads and networks
by Scott Woods 26 Apr '10

26 Apr '10

Hi Boost, Is anyone interested in a framework for asynchronous programming? This is based on the thinking contained in; * SDL (http://www.sdl-forum.org/SDL/index.htm) * Active Objects (http://www.cs.wustl.edu/~schmidt/PDF/Act-Obj.pdf) SDL is the ITU standard for specification of signaling systems. AO is a pattern for concurrent programming. These have been combined to create an environment where objects exchange messages, possibly between threads or even over networks. Objects are created simply; id = create<object_type>(); and once created, anything can be sent to it; send( anything, id ); The framework takes care of threading, network addresses, queueing, encoding and optimal I/O, as appropriate. Threads are equally simple to create; p = start<message_processor>(); and objects can be assigned to specific threads at creation time; id2 = create<object_type>( p ); A network server can be created as simply as; listen = create<socket_listen>(); server = create<socket_server<my_server> >(); service operational( "loopback:5500", server ); send( operational, listen ); Connecting is equally simple; connect = create<socket_connect>(); client = create<my_client>(); service operational( "localhost:5500", client ); send( operational, connect ); >From that point on, client and server can exchange messages using the send primitive without any knowledge that messages are actually travelling across a network. There is in fact no difference between inter-thread messaging and inter-processes (i.e. for the application). Feel free to poke around in the software at; http://groups.google.co.nz/group/pact-serialization/web I am interested in a Boost review of this library but suspect it is too large. It also overlaps with several existing boost components. But good to get further opinions. Cheers, Scott

5 18

[Interprocess] Patch: Various standard-conformance fixes
by Doug Gregor 26 Apr '10

26 Apr '10

The attached patch contains several fixes to improve the interprocess library's conformance to the C++ standard. The fixes are: - Added two missing "template" keywords where the left-hand side of a member access (with ->) is dependent - Removed a "mutable" from a reference type in two places - Removed an extra space from the end of the file name in INCLUDE_BOOST_CONTAINER_MOVE_HPP - Doug

2 1

[Iterators] Iterators and higher-order programming
by Thomas Jordan 25 Apr '10

25 Apr '10

Hi, One of the interesting things about the iterators library, it seems to me, is the additional support it gives for higher-order, functional-style programming. For example, you can use the transform and filter iterators to produce a simple 'Comprehension' object which permits list-comprehension type of syntax to be used, instead of conventional loops (or, e.g., std::transform). I found this quite exciting. I guess I could have come across this nice library sooner if it had been called 'Higher-Order Iterators' or at least co-categorised under the Function Objects and Higher-Order Programming library category. Anyway, is the the intention that we now have a 'computationally-complete' alternative to loops, with which we can write general loops in a more functional, type-ful way? Regards, Tom

2 1

[type_traits] optimized std::copy example wrong?
by Thorsten Ottosen 25 Apr '10

25 Apr '10

Hi John & co, IIRC, std::copy() cannot use memcpy(), since the memory is allowed to overlap. Therefore it can only use memmove(). regards -Thorsten

5 5

[1.43][Serialization] Serialization library seriosly broken
by Hartmut Kaiser 25 Apr '10

25 Apr '10

As I alluded to in my mail yesterday, the serialization library seems to be utterly broken in the current trunk, and unfortunately in the current release branch as well. We're seeing assertions at application initialization time during serialization registration in the file void_cast.hpp, line 225. That is in debug builds. In release builds we're seeing aborts later on. These problems seem to be caused by registering the same type more than once but from different shared libraries. We do not intentionally register the types more than once, even more, we have no control over this. Here is the simplified code causing the multiple registrations from different modules: struct Parameter_impl : ::Par // ::Par is a C-struct { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, const unsigned int version) { ar & lambda; ar & allowedl; // more serialization going on } }; struct Parameter { boost::shared_ptr< Parameter_impl > p; // ... private: friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, const unsigned int version) { ar & p; } }; Note: no BOOST_..._EXPORT or other serialization macros are utilized in this file. This code is included by several files from different shared library modules as it is passed around. It is just including this file into compilation units from different shared libraries what causes the problems described above. All code worked before and still does with V1.42. If this error stays in the release branch the whole Boost release will be unusable for us, causing endless grief in the future. Regards Hartmut --------------- Meet me at BoostCon www.boostcon.com

6 13

[tokenmap] updated perfect hash container in sandbox
by Slawomir Lisznianski 25 Apr '10

25 Apr '10

Changes: * To eliminate a possibility of the algo for finding free slots becoming slower as the load factor increases, the container now keeps track of open slots in unused nodes. This guarantees constant time insertion of new elements at a slight bookkeeping overhead. Consequently, the load-factor has been removed. Pointed by Daniel James. * More STL compliant (iterators, size(), empty(), etc)

4 9

Help with VeecoFTC - exception - cloning_test / msvc-9.0~wm5~stlport5.2
by Emil Dotchevski 24 Apr '10

24 Apr '10

Can someone help me figure out what's going on here: http://tinyurl.com/ybpbsp4 Thanks! Emil Dotchevski Reverge Studios, Inc. http://www.revergestudios.com/reblog/index.php?n=ReCode

3 5

callback or file based construction of CSR graph
by Dan Jiang 24 Apr '10

24 Apr '10

Is there any way for CSR graph to provide a constructor that takes a file stream pointer or a callback functor? If not, how hard is it to add it? I am using edges_are_sorted CSR constuctor. It doubles memory I passed in during constructor calls. My graph is huge (>30 million arcs) and It took >8gb ram to construct my CSR. If I can have callback or file-based construction, I could cut down the memory usage by half. Dan _________________________________________________________________ Hotmail & Messenger. Get them on your phone now. http://go.microsoft.com/?linkid=9724463

2 2

(no subject)
by Dan Jiang 24 Apr '10

24 Apr '10

Is there any way for CSR graph to provide a constructor that takes a file stream pointer or a callback functor? If not, how hard is it to add it? I am using edges_are_sorted CSR constuctor. It doubles memory I passed in during constructor calls. My graph is huge (>30 million arcs) and It took >8gb ram to construct my CSR. If I can have callback or file-based construction, I could cut down the memory usage by half. Dan _________________________________________________________________ Videos that have everyone talking! Now also in HD! http://go.microsoft.com/?linkid=9724465

1 0

[1.43.0] File version.hpp in release branch still says V1.42
by Hartmut Kaiser 24 Apr '10

24 Apr '10

Go figure ... Regards Hartmut --------------- Meet me at BoostCon www.boostcon.com

2 1

[1.43.0] Beta 1 now available from SourceForge
by Beman Dawes 24 Apr '10

24 Apr '10

Boost release 1.43.0 beta 1 is now available from SourceForge See http://sourceforge.net/projects/boost/files/ For details of what's in the release, see http://beta.boost.org/users/news/version_1_43_0. Note that the links to files on this web page are for the final release - use the SourceForge link above to get the beta files. There are known issues with GCC 4.5 and VC++ 2010. These compilers were released a few weeks ago so support is still rough around the edges. Some fixes will make it into the final 1.43.0 release, but full support will not be available until 1.44.0. Please download the beta, give it a try, and report any problems you encounter here. -- The Boost Release Team

3 4

[type_traits] what a about a trait that allows memcpy provided that destructors are not called?
by Thorsten Ottosen 24 Apr '10

24 Apr '10

Hi All, I'm looking into writing a boost::malloc_allocator<T> class. This allocator can also reallocate memory in-place via call's to realloc() which can be a benefit for vector-implementation and all it's functions that grow or shrink the vector. However, it is normally only safe to call memcpy() for objects with a trivial assignment operator. But we may also safely relocate a much larger set of classes, e.g. boost::shared_ptr<T> or std::vector<T>. At first I thought a trait like has_trivial_move_constructor<T> would be enough, but that seems to rule out e.g. smart pointers. The compiler can probably not detect this property, so what about a trait like template< class T > struct is_reallocable; Thoughts? -Thorsten

1 0

[1.43][beta] Fix for MPL under gcc 4.5
by joel falcou 24 Apr '10

24 Apr '10

Can : https://svn.boost.org/trac/boost/ticket/4061 be fixed by merging the attached trivial patch ? gcc 4.5 borks up on a deep MPL internals that prevent some other component to not compile. Thanks -- ___________________________________________ Joel Falcou - Assistant Professor PARALL Team - LRI - Universite Paris Sud XI Tel : (+33)1 69 15 66 35

16 30

[Serialization] Assertion when registering same type from different modules, release branch broken?
by Hartmut Kaiser 24 Apr '10

24 Apr '10

Hi, In the newest version of the serialization library (trunk) I'm seeing assertions in file void_cast.cpp, line 227 during application startup. The reason appears to be that the same type gets registered for serialization from different modules (shared libraries). The very same code worked for us before (and still works for V1.42). I did not check with the release candidate yet, but a comparison of the file void_cast.cpp in trunk and release branch lets me assume the error is there as well (both files are identical). I will try to recompile everything on the release branch over the weekend. If the same behavior shows up in the release branch I'd consider the serialization library seriously broken. The problem is, that I really don't know how to avoid this multiple registration from different modules as the type in question is defined in a header file (no EXPORT macro) which is being used from different modules. I don't think this is something really special. Any ideas what we could do? What's the rationale of changing the registration behavior in the serialization library? Regards Hartmut --------------- Meet me at BoostCon www.boostcon.com

1 0

Request to merge r61245 into release branch
by Jeremiah Willcock 23 Apr '10

23 Apr '10

Is it too late for this one? It fixes one warning (class vs. struct) and several errors for VC++ 10 (ambiguous begin and end calls). -- Jeremiah Willcock

2 1

[range] Merge to fix defects
by Neil Groves 23 Apr '10

23 Apr '10

Boosters, I have had a commit on the trunk to resolve some defects that I noticed by carefully reviewing all of my new code. There is one serious defect in the istream_range which on the release branch is useless. Of course, this is merely a lack of a new feature rather than a regression but I would like to have it working for the proper release. There are also some changes to return types to address the concern brought up in the review about inconsistent return types of various functions. Both of these fixes have been tested on the trunk for many days. I recently reverted a host of other changes on the trunk that are unrelated to these issues that would increase risk. The recently reduced changeset on the trunk is also passing regression tests. Thanks in anticipation, Neil Groves

2 3

[GIL] [Patch] References cannot be mutable
by Doug Gregor 23 Apr '10

23 Apr '10

boost/gil/extension/dynamic_image/apply_operation_base.hpp has two references marked "mutable", at lines 141 and 157, but references cannot be mutable. This patch removes the "mutable" keyword from both cases. - Doug

1 0

[any] allocator support?
by Kenny Riddile 23 Apr '10

23 Apr '10

This may have been brought up before, but I have a case where I need to use boost::any with types that require custom allocation. As far as I can tell, this isn't currently possible. Is there a reason boost::any doesn't allow specifying an allocator? std::allocator would obviously be the default. Has anyone maybe done this themselves already and be willing to share?

1 0

[geometry] few questions
by Janek Kozicki 23 Apr '10

23 Apr '10

Hello, By looking at the documentation I can see that Geometry library is great for 2D polygons. I am working in 3D, so I wanted to ask about what I can, and what I can't. Is it possible to define 3D shapes, calculate their intersections, unions and so on? What about centroids and second (geometrical) moments of inertia? Does the geometry deal with spheres, or maybe other classical 3D geometries, like ellipsoid, torus or cone? What about the infinite geometries (like infinite cone)? Or infinitely long line? Also I wanted to ask about 3D points regarded as vectors. Can I add and subtract them? What about 3D vector cross product. What about dot product? I guess that I need to write that 'very old formula' function myself? In fact, since you are well into the subject, perhaps you could tell me if there are anywhere in boost a vector dot product and cross product implemented? Last but not least - I am doing frequently a 3D rotation of 3D points using quaternions for that (this could work with rotation matrices too). I suppose that I would need to implement this myself to "link" boost::quaternion with boost::geometry::point3d. please enlighten me and thank you very much for your time -- Janek Kozicki http://janek.kozicki.pl/ |

3 4

BOOST_EXPLICIT_OPERATOR_BOOL macro?
by Christian Holmquist 23 Apr '10

23 Apr '10

Hi, IIUC, c+0x allows the safe bool idiom to be implemented as explicit operator bool() const { ...} but in c+03 emulating that behaviour is a bit messy. There appears to be a solution for c+03 in boost/smart_ptr/detail/operator_bool.hpp, but it's targeted for the smart_ptr classes only. Would it be possible to extract operator_bool.hpp into a reusable macro, something like: #include <boost/utility/operator_bool.hpp> struct some { BOOST_EXPLICIT_OPERATOR_BOOL( m_value != 0; ) int m_value; }; What do you think? Cheers, Christian

3 5

[range] new adaptor for iterating over equal_ranges
by Thorsten Ottosen 23 Apr '10

23 Apr '10

Hi Neil & others, I think it could be useful with a range adaptor of the following kind to iterate over equal_ranges in a multimap. E.g. SomeMultiMap map = ...; for( auto rng : map | boost::adaptors::equal_ranged ) for( auto iter : rng ) { // do something with the objects with equal keys } The resulting range would probably be bidirectional with value_type std::pair<boost::range_iterator<SomeMultiMap>::type,...>. Thoughts? -Thorsten

2 1

Fwd: Re: [interprocess][move] enum template argument fails to be a valid base class
by Ion Gaztañaga 23 Apr '10

23 Apr '10

>i think I spoke too fast, is_enum needs compiler intrinsics according >to >boost documentation, so maybe this approach is not very useful... Forget it. I was talking about is_union instead. God, the lack of coffee is killing me ;-) Best, Ion

2 4

[1.43.0] Beta 1 Release Candidates available
by Beman Dawes 23 Apr '10

23 Apr '10

The 1.43.0 Beta 1 Release Candidate files are available at http://boost.cowic.de/rc/ Before these get pushed out to SourceForge, we like to get verification that they download and build OK on at least Linux and Windows. So please download one of these for your favorite platform, try a build, and report success or failure here. Thanks, --Beman

5 9