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

Re: [Boost-users] [boost] [review][constrained_value] Review ofConstrainedValueLibrary begins today
by Jesse Perla 10 Dec '08

10 Dec '08

>This too is already solved if any predicate is acceptable: just use a >predicate functor that forwards via a pointer or reference to another >global predicate. Or (somewhat less efficient but easier) use the >dynamic boost::function predicate that is already included and >boost::bind/lambda/phoenix to a method on this global object. Gordon: I absolutely agree that if this library was purely for bounds testing, managing invariants, etc, that predicates are sufficient. But for high performance numerics, that is not what would happen. First of all, you would turn off automatic testing at runtime. But one of the key value of this library, in both release/debug, for those examples is to get information about the bounds and intervals from the type itself for generic algorithms. A predicate as a function can test assignment, but it can't give back structured data for these algorithms. If you can show me how the testing predicate can be easily queried within this framework for the type metadata itself(e.g. a list of intervals), then I will shut up. I really think people need to think through this alternative use case of this powerful library. And Robert: There are a lot of discussions on the boost mailing list (which I can't post on) about NaN and others for numeric_limits. In virtually every case with high performance computing, you would turn off predicate testing on assignment for release builds because it is too expensive. Then you would, judiciously, test inclusion of the value. So the model here is that if bounded_float is a subset of float, a value in this class it is always in the float superset, but I can test inclusion within the other subset that I have defined. Here, predicates work fine as long as there is a manual way to test set inclusion. And numeric libraries frequently use NaN, infinity, etc. as signals for algorithms. So these values should be associated with the superset in the model as a default approach. So what I am saying is that people should be given the option (as a preprocessor instruction is sufficient) to have the value as always within the bounds vs. considering this as a "set" which they can then test inclusion at their own liesure. This is a pragmatic approach to ensure consistency of this library with existing generic numerical algorithms, which is just as important for the numerics as solving the floating point problem. Also, I am not aware of the structure of the numeric_limits, but I doubt these are written to be virtual/subclassed due to efficiency. What this means is that you would probably have to write your own traits class if we don't make this automatic, and you can't pick or choose to override existing tratis. And these numeric traits are impossible for normal developers to write since they are platform specific. Last, and this is not something I know enough about because I am not a numeric analyst, but the epsilon method discussed in the mailing list may have a flaw for numeric work. What we effectively are doing here is creating an epsilon neighborhood for testing equality. As we know from set theory/topology, this does not create a weak ordering because it fails transitivity. Now, one might say that this is irrelevant because it is a reasonable approximation for pragmatic reasons, but you run into a problem in numerics. The reason is that many numerical algorithms are already testing within an epsilon neighborhood for convergence of algorithms, derivatives, etc. So we need to be REALLY careful that this epsilon ball is well within those balls, or we may seriously mess up algorithms depending on a comparison operator.... This may or may not be a problem, but I would pose it on the boost list to ensure that people are thinking about it. One solution may be to rely on the numeric_limits::epsilon trait that would always be available, would be platform specific, and I believe is the small possible neighborhood of a point. Whether this is the maximum amount of truncation possible, I do not know. -Jesse

3 2

signals/slots and boost::functions
by david.weber＠l-3com.com 10 Dec '08

10 Dec '08

I have a class, where I want callers to be able to register/deregister signals/slots. They will pass in a function pointer, and we'll hook it up to the signal, and manage the connection object. Part of the management, is I need to map the function pointer to the actual connection object. This is where the trouble comes in. Since a boost::function doesn't have an operator <, the simplest way is to just map it to an unsigned long long, and store the raw address of the underlying function. So, my connect function: typedef VSSP (*TargetType) ( std::string, EncodingEnum); typedef boost::function < VSSP ( std::string, EncodingEnum) > CreateTypeFunc; bool ElementBase::connectCreateTypeSignal(const CreateTypeFunc& slot) { LOG4CXX_DEBUG(logger_, "connectCreateTypeSignal in"); bool retValue = false; unsigned long long slotAddr = (unsigned long long)slot.target< TargetType >(); LOG4CXX_DEBUG(logger_, "*** slotAddr = " + (boost::format("%08X") % slotAddr).str()); <snip> } One of my checks, is to ensure that you don't connect the same thing multiple times. So, I have a test case that does this, and expects it to bomb out: VSSP getTypeSlot(std::string fqn, EncodingEnum encoding) { <snip> return VSSP; } BOOST_FIXTURE_TEST_CASE(ElementBase_connectCreateType, ElementBaseTestFixture) { BOOST_CHECK(true == ebp->connectCreateTypeSignal(getTypeSlot)); BOOST_CHECK(false == ebp->connectCreateTypeSignal(getTypeSlot)); } My debug printouts, are shoing unexpected behavior. The address that I am getting from target is different every time it is called, even though I'm calling with the same function (getTypeSlot). I would expect this if I was getting the address of the function object, because I'm creating one implicitly through the call, but not dereferencing via target(), unless I'm misunderstanding what target() does. I just need the stupid address of the function passed in. Help!! TIA --dw

3 5

Eigenvalues
by r89 10 Dec '08

10 Dec '08

Hi, It seems like Boost doesn't support eigen-analysis, is that true? I've been using VXL/VNL for eigenanalysis, but my project must get rid of dependencies on VXL/VNL, so I'm wondering what would be the easiest transition? By the way, the matrices I'm dealing with are of size 2x2 and 3x3. Thanks! -- View this message in context: http://www.nabble.com/Eigenvalues-tp20941764p20941764.html Sent from the Boost - Users mailing list archive at Nabble.com.

4 3

[BGL] passing a graph from one function to another
by r89 10 Dec '08

10 Dec '08

Hi all, I have a graph defined as follows (the same as in the Dijkstra shortest path example code): typedef adjacency_list < listS, vecS, directedS, no_property, property < edge_weight_t, float > > graph_t; The following works fine: graph_t myGraph(array, edge_array + num_arcs, weights, num_nodes); However, now I want to do the above line inside a function (say, fun_A), and I want another function (say, fun_B) to have access to "myGraph". How should I do that? I wish I could do something like this: graph_t myGraph; fun_A(graph_t &myGraph) { myGraph(array, edge_array + num_arcs, weights, num_nodes); } fun_B(graph_t &myGraph) { // using myGraph to run some graph algorithm... } but it doesn't work. Any suggestions? Thanks! -- View this message in context: http://www.nabble.com/-BGL--passing-a-graph-from-one-function-to-another-tp… Sent from the Boost - Users mailing list archive at Nabble.com.

5 12

[bind/lambda] Matching signature constness?
by Robert Jones 10 Dec '08

10 Dec '08

Here's a bit of code (and my apologies for it's length, but I'm unsure which bit is the failure point) #include <cstddef> #include "boost/lambda/lambda.hpp" #include "boost/lambda/bind.hpp" #include "boost/function.hpp" #include "boost/type_traits.hpp" namespace { template < bool > struct cmp_detail { template < typename T > static bool cmp( const T & lhs, const T & rhs ) { return lhs != rhs; } }; template < > struct cmp_detail< true > { template < typename T, std :: size_t sz > static bool cmp( const T ( & lhs )[ sz ], const T ( & rhs )[ sz ] ) { using boost :: lambda :: _1; using boost :: lambda :: bind; using boost :: lambda :: var; boost :: function< bool( unsigned ) > cmp_element = bind( cmp_detail< boost :: is_array< T > :: value > :: cmp, var( lhs )[ _1 ], var( rhs )[ _1 ] ); for ( unsigned i = 0; i != sz; ++ i ) if ( cmp_element( i ) ) // if ( cmp_detail< boost :: is_array< T > :: value > :: cmp( lhs[ i ], rhs[ i ] ) ) return true; return false; } }; template < typename T > bool cmp( const T & lhs, const T & rhs ) { return cmp_detail< boost :: is_array< T > :: value > :: cmp( lhs, rhs ); } template < typename T, std :: size_t sz > struct Block { typedef T type[ sz ][ sz ]; }; } int main( ) { Block< int, 16 > :: type a; Block< int, 16 > :: type b; cmp( a, b ); } The problem is the bind call in cmp_detail<true>::cmp(...). Error is this block.cpp: In static member function `static bool <unnamed>::cmp_detail< true>::cmp(const T (&)[sz], const T (&)[sz]) [w ith T = int[16], unsigned int sz = 16u]': block.cpp:45: instantiated from `bool <unnamed>::cmp(const T&, const T&) [with T = int[16][16]]' block.cpp:61: instantiated from here block.cpp:32: error: no matching function for call to `bind(<unknown type>, const boost::lambda::lambda_functor<boost::l ambda::lambda_functor_base<boost::lambda::other_action<boost::lambda::subscript_action>, boost::tuples::tuple<boost::lam bda::lambda_functor<boost::lambda::identity<const int (&)[16][16]> >, boost::lambda::lambda_functor<boost::lambda::place holder<1> >, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boo st::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type> > >, const boost::l ambda::lambda_functor<boost::lambda::lambda_functor_base<boost::lambda::other_action<boost::lambda::subscript_action>, b oost::tuples::tuple<boost::lambda::lambda_functor<boost::lambda::identity<const int (&)[16][16]> >, boost::lambda::lambd a_functor<boost::lambda::placeholder<1> >, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples:: null_type> > >)' I think perhaps the var() calls are not preserving constness quite right? Any help would be welcome. Thanks. - Rob.

2 3

[Review] UUID library (mini-)review starts today, November 23rd
by Hartmut Kaiser 10 Dec '08

10 Dec '08

Hi all, The mini-review of Andy Tompkins UUID library starts today, November 23rd 2008, and will end on November 30th. I really hope to see your vote and your participation in the discussions on the Boost mailing lists! The library can be downloaded from the vault here (it's the file uuid_v13.zip in the topmost directory): http://tinyurl.com/6xx28b The initial review of the UUID library ended with a provisional acceptance (read here: http://article.gmane.org/gmane.comp.lib.boost.user/27774/) This mini review is meant to allow for a final decision after looking at the changed parts of the library. Here is a list of things fixed and changed: - fixed the licensing issues revealed by the initial review - fixed the security vulnerability revealed by the initial review - renamed to uuid, moved all classes, functions, etc. to namespace boost::uuids - implemented sha1 hash function (thus no license problem) - new class basic_uuid_generator to create random-based uuids. The random number generator is no longer hard coded. It can use any random number generator, default is boost::mt19937 - implemented a good seed function for random number generators - all functions are now reentrant, all classes are as thread safe as an int and the library is no longer dependent on Boost.Thread --------------------------------------------------- Please always state in your review, whether you think the library should be accepted as a Boost library! Additionally please consider giving feedback on the following general topics: - What is your evaluation of the design? - What is your evaluation of the implementation? - What is your evaluation of the documentation? - What is your evaluation of the potential usefulness of the library? - Did you try to use the library? With what compiler? Did you have any problems? - How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? - Are you knowledgeable about the problem domain? Regards Hartmut Review Manager

14 39

Dplicate emails
by Richard Howells 10 Dec '08

10 Dec '08

Since a couple days back I seem to be getting every email twice. It is coincident with me repaving my computer, but it does NOT happen to my individual emails nor to my other mailing list subscriptions. Is it just me? Cheers, Richard richard(a)dynamisys.co.uk www.dynamisys.co.uk L: (+44) (0)1793 731225 M: (+44) (0)7732 971786

2 1

[boost.build] multiple definitions of symbols at link since revision 49800.
by Nigel Dyer 10 Dec '08

10 Dec '08

Hello I am using the trunk release of boost, and started having problems with multiple definitions of symbols during linking (see example linker output below). I can cure the problem by removing the change to tools\build\v2\build\generators.jam introduced in revision 49800. It only seems to occur if the following headers: #include <boost/test/floating_point_comparison.hpp> #include <boost/lambda/lambda.hpp> #include <boost/lambda/bind.hpp> are included in precompiled headers which are included in cpp files that are created by the antlr parser generator. It occurs on a compilation with darwin, but not msvc. It is unclear whether the generator.jam change has identified a waekness in the boost libraries, or there is a problem with the revision 49800 change that only surfaces in obscure circumstances, or the change has highlighted a problem in my code. It can be fixed by removing the includes from the precompiled header. I feel this should not be necessary As a boost newbie it is unclear whether I should I change my code or raise a bug report. I would be grateful for suggestions. Nigel Example linker failures. /usr/libexec/gcc/i686-apple-darwin8/4.0.1/ld: multiple definitions of symbol boost::test_tools::(anonymous namespace)::check_is_close /Volumes/RAID1/xraid/data/bifa/Dev/Projects/Bio/bin/Bio/darwin-4.0.1/release/link-static/libbio.a(BiobaseLexer.o) definition of boost::test_tools::(anonymous namespace)::check_is_close in section (__DATA,__const) /Volumes/RAID1/xraid/data/bifa/Dev/Projects/Bio/bin/Bio/darwin-4.0.1/release/link-static/libbio.a(ArrayLexer.o) definition of boost::test_tools::(anonymous namespace)::check_is_close in section (__DATA,__const)

4 3

[BGL] iterator/descriptor invalidation
by Julius Ziegler 10 Dec '08

10 Dec '08

Dear list, the documentation of adjacency_list states, that a call to add_vertex() does not invalidate any iterators nor descriptors. However, following code segfaults, because retrieving vertex descriptors from an edge_iterator via source() and target() does not work (example should compile as is): #include <iostream> #include <boost/graph/adjacency_list.hpp> using namespace boost; typedef adjacency_list<listS, vecS, directedS > Graph; int main() { Graph g(3); add_edge( 0, 1, g ); add_edge( 1, 2, g ); add_edge( 2, 0, g ); Graph::edge_iterator ei, ei_end; for( tie( ei, ei_end ) = edges( g ); ei != ei_end; ++ei ) { std::cout << source( *ei, g ) << " " << target( *ei, g ) << "\n"; add_vertex( g ); // <- in my real application, vertex gets some // properties that depend on source(...) and // target(...) above } } Is it the case that an edge_iterator does not store vertex_descriptors (that should survive reallocation of the vertex storage), but iterators to the vertex storage? What is the proper way to accomplish something like this? My present workaround is to postpone all add_vertex() calls after the loop, but this costs some storage, and with my graph scaling bigger and bigger, that begins to hurt. Any help and explanation is really appreciated! Regards, Julius

3 18

Re: [Boost-users] [boost] [review][constrained_value] Review of ConstrainedValueLibrary begins today
by jesseperla＠gmail.com 10 Dec '08

10 Dec '08

Robert: >First, I hope that you are aware of the issues with using floating point types >with this library. If not, please read the rationale section of the >documentation and the discussion in the related thread on the developers' list. Yup, been following. You did a good job of educating me on a previous post... I think I understand the issue now, but don't really have much of an option. As far as I can undestand from the discussions, this is a general problem with equality operations with floating points and has nothing to do with your library. If this is true, I would change the "rationale" section to say that the mess is not your fault and it isn'ta library problem and that you need to be just as careful as you would normally. >The constraint could look like this (written out of my head, not tested): If this is possible, I am very happy. It looks like it is. The only other things that I would say need to be thought through in the design is unions of constraints instead of just intersections of them. For example, I was checking this out a while back, and it manages the sets of intervals/etc. well: http://www.herold-faulhaber.de/itl/ I would wonder if formalizing intervals in this library (or using another) is a good way to deal with different bounds... but that is beyond my level of competence. > Why not just use numeric_limits<risk_aversion::value_type>? 1) Because I was too stupid to think of that, and 2) Because I want to be able to write generic functions such as: template<typename T> bool is_NaN(T& t) { return (t == std::numeric_limits< T >::quiet_NaN()); } Again, my idea is that I want to be able to throw around elements of constrained types to generic functions and all of the standard types of things I would do with the superset type. >Actually, only the mutating operators (++, --, =, += etc.) are overloaded. For >I don't think this is needed. If you write: > constrained<int> x, y, z; > z = x + y; >Then simply the + operator for int is used. Even better. As long as it works for all of the auto conversions between intrinsic types we are also in business. eg constrained<int> x, y, z; double c; z = c * x + y; Thanks Robert. Great library, time for me to submit a glowing review.

4 4