- Boost - lists.boost.org

[metagraph] mpl_graph and MSM
by Gordon Woodhull 14 May '10

14 May '10

Friends, I'm pleased to announce that I'll soon be submitting mpl_graph for review. This is an implementation of the BGL concepts for compile time metaprogramming. Currently it supports incidence list, adjacency list, depth first search, and breadth first search. It is available in the sandbox under metagraph. I'll write some documentation soon; for now consult BGL's concepts, which are almost the same, just s/()/ typename <>::type/ ;-) It's pretty small - 825 lines of code so far. Standing on the shoulders of MPL... I've also provided a minimal example of running depth first search on the canonical MSM CD Player example. See sandbox/metagraph/libs/ metagraph/example/msm_adaptor.cpp Christophe, thanks for prodding me to get this out the door in your talk here at BoostCon! It sounds like we should write a connected components algorithm to get the features you are looking for, which will be easy with the right dfs visitor. Running DFS does cost in compile time. I haven't yet attempted to profile or optimize the template instantiations, and would appreciate any advice. On this example, it increases the compile time from 3.775 to 5.634 seconds on gcc 4.4. Since the metadata graph structure is lazy (you pay approximately for each graph concept that you exercise), construction of the graph is only 0.2s out of the 1.85s. Interestingly, running BFS in addition to DFS on the same graph adds less than 0.2s -- this makes sense because the big expense should be generating the mpl::maps which support vertex and edge lookup. Here is the test I ran. (I took the median time from each DFS level.) for((dfs=0;dfs<6;++dfs)); do echo; echo DFS=$dfs, 5 runs; for((i=0;i<5;++i)); do time g++ -D DO_DFS=$dfs -I ../../.. -I ~/src/MSMv2.10 -I ~/src/boost- trunk msm_adaptor.cpp -o msm_adaptor; done ; done The good news is that it takes less than 25 lines of code to turn the transition table into a graph and run DFS on it. Thanks to everyone for a truly awesome BoostCon! Gordon P.S. mpl_graph is part of a larger effort to generate patterns of objects from graph metadata. (Metagraph to me means "graph of graphs".) Would people rather I submitted mpl_graph separately, or keep it as a sublibrary of metagraph? (The full metagraph is still months or years away.) P.P.S. Gurtovoy/Abrahams, is the name OK? It's meant to emphasize that it's a simple extension of MPL into graphs.

2 2

[GIL] How do you use GIL?
by Lubomir Bourdev 14 May '10

14 May '10

If you are using GIL in your project and don’t mind being quoted, could you please email me your name, your project/company or university and any optional comments? We will be giving a tutorial for Boost.GIL at the CVPR 2010 conference in San Francisco on June 13. CVPR is an international conference in computer vision attended by many people from both industry and academia. This is an opportunity to promote the boost libraries in general. Thanks, Lubomir

1 0

Fwd: [MPI] Test Processor Counts
by Belcourt, Kenneth 14 May '10

14 May '10

Sent to the users list by mistake. Begin forwarded message: > From: "Belcourt, Kenneth" <kbelco(a)sandia.gov> > Date: May 14, 2010 11:03:23 AM MDT > To: "boost-users(a)lists.boost.org" <boost-users(a)lists.boost.org> > Subject: [MPI] Test Processor Counts > > Hi, > > While running the trunk tests, we've found that occasionally the > system load average jumps up over 100 and occasionally hits 200. As > you can imagine for 8 to 32 core systems, this adversely impacts users > on those systems. > > The issue is that I'm running the trunk tests in parallel, usually -j8 > to -j32 and when the MPI tests run, we run into the high load > problem. Some of the MPI tests run up to 17 processors so the > combination of parallel testing and large processor count tests cause > this issue. Now the high load average may last for only 5 to 15 > minutes or so but this is enough to trigger a sysadmin question asking > what I'm up. > > I'm looking for a solution that wouldn't require significantly > reducing testing throughput. Possible solutions might include > serializing the execution of the MPI tests using something like the > link-semaphore in Boost.Build, reducing the processor counts of the > MPI tests, only running the larger tests on unchecked or unknown MPI > implementations, setting up an MPI machine file to run the MPI tests > off of the local system that built the executable, etc... > > Any other ideas on how to manage this? > > Thanks. > > -- Noel >

1 0

[regex] Last VC6 version
by Roger Orr 13 May '10

13 May '10

The page http://www.boost.org/doc/libs/1_43_0/libs/regex/doc/html/boost_regex/install .html still states that VC6 can be used to build regex, but this no longer appears to be actually the case. Does anyone know which was the last version of boost regex to support VC6? Regards, Roger.

2 2

Patch to value_initialized
by Christopher Jefferson 13 May '10

13 May '10

I just wanted to make people away of a suggested patch to value_initialized, explicitly adding the default constructor, copy constructor and operator=, given in https://svn.boost.org/trac/boost/ticket/4213 . The default constructor is necessary for clang and comeau (in strict mode) to allow const value_initialized values without an explicit value. I believe this is correct behaviour, but other compilers are more lack. I believe this patch is small and should work fine, but I know that this has been a very sensitive bit of code in the past, prone to kicking up minor compiler problems. My preferred option would be to try submitting this patch to trunk and see if any tester kicks up a fuss. If any do, then I am happy to introduce some macros.

2 2

trac database locked
by Kim Barrett 13 May '10

13 May '10

I'm trying to submit a patch via trac and get the error "OperationalError: database is locked"

2 2

[interprocess] flat_set/flat_map should add front(), back(), operator[], pop_back()
by Thorsten Ottosen 13 May '10

13 May '10

Hi Ion, When using e.g. a flat_set as a priority queue, it is annoying not to have back() and pop_back() available. As for operator[](), then of couse cont.begin()[n] works, but this is somewhat less pretty compared to just cont[n]. Therefore I would really like to see all the above functions added to the interface. kind regards Thorsten

2 2

[LA] vector_traits for setters
by Daniel Pfeifer 13 May '10

13 May '10

Hi Emil and others, Qt 4.6 brought support for some vector and matrix types. While it is easy to define matrix_traits for QGenericMatrix and QMatrix4x4, i am struggling with the vector_traits for QVector2D, QVector3D and QVector4D. The problem is that Qt's vector types have setters and no non-const-ref getters. Is there a plan for (Boost).LA to support types with setters? cheers, Daniel Documentation for Qt's vector and matrix types can be found here: http://doc.qt.nokia.com/4.6/qmatrix4x4-members.html http://doc.qt.nokia.com/4.6/qgenericmatrix-members.html http://doc.qt.nokia.com/4.6/qvector2d-members.html http://doc.qt.nokia.com/4.6/qvector3d-members.html http://doc.qt.nokia.com/4.6/qvector4d-members.html

4 10

[exception] warnings from BOOST_THROW_EXCEPTION
by Steven Watanabe 13 May '10

13 May '10

AMDG The attached patch fixes warnings like ..\..\../boost/property_tree/string_path.hpp:216: error: control reaches end of non-void function Okay to commit? In Christ, Steven Watanabe

3 3

[svn sandbox access] svn sandbox access requested via correct method but no reply
by Dan Walters 12 May '10

12 May '10

Hi all, I emailed boost-owner(a)lists.boost.org on 27 April 2010 23:32 requesting sandbox svn access but am yet to recv a reply. After asking in #boost, DHulme recommended that I post here mentioning this. Thanks for your time, Dan

3 3