November 2008 - Boost - lists.boost.org

Group libraries by topic in the documentation
by Kevin Sopp 11 Nov '08

11 Nov '08

There are 92 entries in the library overview now (http://www.boost.org/doc/libs) I believe this is quite intimidating for a new user. I have come up with a scheme of grouping things (I don't claim it is perfect and I put libraries that I wasn't sure about into "Others"). Each library should only appear once even if that one category doesn't describe the library's purpose perfectly. The "Others" section could be subdivided further, i.e. Concurrency and Time categories come to mind but I didn't deem that necessary. Additionally, it would be nice to 1) indicate header-only status via a preceding (+) symbol and a library-only status via (-). This can be combined into (+/-) for libraries that support both. 2) indicate deprecated library status via "deprecated, use XXX instead". 3) indicate expected learning curve via color (green, yellow, red) where red is intended for something like Proto and green for Any. Yellow could be used for a library like Program Options. Containers----------------------- Any Array Bimap Circular Buffer Dynamic Bitset Intrusive Multi-Array Multi-Index Pointer Container Tuple Unordered Variant Functional Programming----------- Bind Functional Lambda Member Function I/O------------------------------ Filesystem Format IO State Savers Iostreams Serialization Math----------------------------- Accumulators Disjoint Sets Interval Math Math Common Factor Math Octonion Math Quaternion Math/Special Functions Math/Statistical Distributions Random Rational uBLAS Memory Management---------------- Pool Smart Ptr Metaprogramming------------------ Call Traits Compressed Pair Concept Check Enable If Function Types Fusion MPL Operators Parameter Preprocessor Property Map Proto Statechart Static Assert Type Traits Typeof Networking----------------------- Asio Text processing------------------ Regex Spirit String Algo Tokenizer Xpressive Others--------------------------- Assign CRC Conversion Date Time Exception Foreach Function Functional/Hash GIL Graph In Place Factory, Typed In Place Factory Interprocess Iterators Min-Max MPI Numeric Conversion Optional Program Options Python Range Ref Signals System Test TR1 Thread Timer Units Utility Value Initialized Tribool Wave Boost Internal------------------- Compatibility Config Integer

5 5

[Website] Navigation and discoverability
by David Abrahams 11 Nov '08

11 Nov '08

Hi all, I still have a hard time finding my way around the website since it was restructured. I'm having at least two problems: 1. I can pretty much guarantee that manually changing a URL in the address bar of my browser will fail, I think because the URLs are deep and relatively unforgiving. Unfortunately I have little or no insight into what would fix this problem. 2. Many files have version numbers in their URLs, and there doesn't seem to be any way to link to "whatever the latest is." Also, once we have that shouldn't we try to encourage search engines to prioritize that URL as opposed to any of the many near-duplicates (not that I know how)? 3. I can't get a good feeling for what's on the site and how it's structured without hopping around all over the place. For example, if I'm looking for mailing lists, do I want "community" or "support?" The answer isn't immediately obvious. http://www.boostpro.com has drop-down menus that are entirely done with highly portable CSS; I got the base code from http://www.cssplay.co.uk/menus/final_drop.html. Something like that would be an immense help in finding what I'm looking for. Thoughts? -- Dave Abrahams BoostPro Computing http://www.boostpro.com

3 2

BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS causes build to fail
by Steve M. Robbins 11 Nov '08

11 Nov '08

Hi, If the symbol BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS is defined (see <boost/math/tools/config.hpp>) then the build fails [1] as follows: libs/math/build/has_long_double_support.cpp:9:2: error: #error "long double support is not supported by Boost.Math on this Plaform: the long double version of the TR1 library will not be built." ...failed gcc.compile.c++ bin.v2/libs/math/build/has_long_double_support.test/gcc-4.3.2/release/debug-symbols-on/link-static/has_long_double_support.o... and several others, ending with: ...failed updating 8 targets... ...skipped 368 targets... ...updated 2074 targets... make: *** [build-stamp] Error 1 The failure is caused by an #error directive in has_long_double_support.cpp, suggesting this behaviour is deliberate. How can this be made to be non-fatal? Thanks, -Steve P.S. The behaviour is easy to reproduce: just stick #define BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS into config.hpp. [1] http://buildd.debian.org/fetch.cgi?&pkg=boost1.36&ver=1.36.0-3&arch=hppa&st…

1 0

[threadpool] new version - fork/join (recursive/fibers)
by k-oli＠gmx.de 11 Nov '08

11 Nov '08

Hello, v13: http://www.boostpro.com/vault/index.php?action=downloadfile&filename=boost-… after the discussion yesterday I've modified the code. Now the pool can be configured (second template argument of tp::pool) to use recursive invokation of tasks or fibers in order to enable fork/join semantics. I've found that the recursive invokation is more than two times faster than using fibers (see examples recursive_fibonacci_recursive_task.cpp and recursive_fibonacci_with_fibers.cpp). The code using fibers sometimes raises an assertion in ~mutex() (used in work-stealing queue). For this reason and following the discussions I think I'll remove the support for fibers from threadpool. regards, OPliver

6 24

[signals2][review] The review period of the signals2 library is over
by Stjepan Rajko 11 Nov '08

11 Nov '08

Hello all, The review period of the signals2 library finishes today, November 10th. I would like to thank the author, Frank Mori Hess, for preparing the library for review and for diligently answering questions, and all those that have participated in the review for their thoughtful comments and reviews. A few more reviews will be coming late - if anyone that hasn't contacted me yet would like to submit a review in the following few days please let me know so I know to wait for your review. I will start working on summarizing the content of the reviews and comments submitted so far, and will post a report within a week. Kind regards, Stjepan

1 0

[chrono] Help need for Linux implementation
by Beman Dawes 11 Nov '08

11 Nov '08

I've got a boostified version of the C++0x <chrono> header working on Windows, but need help for the Linux implementation. I wrote a test program (see below). It compiles OK but gets a link failure: /tmp/ccmT80bD.o: In function `clk()': test.cpp:(.text+0x192): undefined reference to `clock_gettime' collect2: ld returned 1 exit status What am I doing wrong? The command I'm using is simply "g++ test.cpp". Ubuntu 8.04 with all updates applied. Thanks, --Beman ------------------------------- #include <time.h> #include <stdlib.h> #include <errno.h> #include <iostream> unsigned long long clk() { timespec ts; if ( clock_gettime( CLOCK_MONOTONIC, &ts ) ) { std::cout << "errno " << errno << "\n"; exit(0); } return (unsigned long long)ts.tv_sec * 1000000000 + ts.tv_nsec; } int main() { std::cout << clk() << " nanoseconds\n"; return 0; }

2 2

[signals2][review] The review of the signals2 library is scheduled to end on Monday, November 10th
by Stjepan Rajko 10 Nov '08

10 Nov '08

Hello, The review of the signals2 library is scheduled to end on Monday, November 10th. We have had some good discussion so far and a couple of reviews. I hope the remaining few days will provide enough time to others that were planning on submitting a review. If anyone needs any more discussion time or time to write the review, please let me know (so we can see about extending the review period, or so that I know to wait for your review longer). Kind regards, Stjepan

4 3

Boost.System support for Linux 2.4 kernel?
by J. van der Wulp 10 Nov '08

10 Nov '08

Hello Everyone, Is it possible to add compatibility for the Linux 2.4 kernel to the system library? The file boost/system/linux_error.hpp refers to a few macros/constants that are not defined for the linux 2.4 kernel. The exact names of the symbols are: EKEYEXPIRED, EKEYREJECTED, EKEYREVOKED ENOKEY. Is Linux 2.4 a supported platform for Boost? Thanks, Jeroen

3 2

Library names in system layout on linux
by Vladimir Prus 10 Nov '08

10 Nov '08

Hi, we had discussion about the library names that are produced on Linux, with --layout system: http://thread.gmane.org/gmane.comp.lib.boost.build/19529 Two person who package Boost for different Linuxes participated, and I think we've agreed on some changes that will make them, and hopefully linux users, a bit happier. Now, with --layout=system, the libraries are named like this: libboost_filesystem_mt.so The proposal is to: 1. Drop 'mt', since most libraries are built with MT on Linux, and don't use such special decorations. 2. Add version number at end, to comply with standard packaging. So, we'd create: libboost_filesystem.so.1.38.0 together with a symlink to it, named libboost_filesystem.so I suspect some folks packaging Boost might not be reading Boost.Build list, so I wanted to mention this plan here. Anybody has objections? - Volodya

7 14

[serialization] rvalues and const-correctness?
by David Abrahams 10 Nov '08

10 Nov '08

Consider: struct X { int f(); template <class Archive> void save(Archive& ar, unsigned) { ar << f(); // can't bind non-const reference to temporary } BOOST_SERIALIZATION_SPLIT_MEMBER() }; This error seems silly to me. Couldn't we add this overload to the archives, or even simply replace the existing operator with this one? template<class T> Archive & operator<<(T const & t){ // ^^^^^------------- Note this->This()->save_override(t, 0); return * this->This(); } TIA, -- Dave Abrahams BoostPro Computing http://www.boostpro.com

3 11