- Boost - lists.boost.org

[fit] Mini Review (consider 1/2 or 1/4 weight)
by Artyom Beilis 21 Mar '16

21 Mar '16

Disclaimer: please consider my review as partial if at all because of the way I did it (see notes below) > - Whether you believe the library should be accepted into Boost No, I don't think so. I don't feel that library adds any real value to boost: being useful or in other words significantly simplifying rather straight forward tasks can't be done However I understand that there are some cases with template meta-programming that can make library useful shown there: http://article.gmane.org/gmane.comp.lib.boost.devel/265453 So I do think it another major review after significant redesign can be done. > - Your name Artyom Beilis > - Your knowledge of the problem domain. I worked a lot with lambda expressions and various callbacks also mostly in dynamic manner rather implementing functional programming. > - What is your evaluation of the library's: > * Design I did very brief analysis of design. What concerns me is wide use of static objects and declaration of "static functions". There is significant difference between static struct foo_class { void operator()() const;... } foo; and inline void foo() { } Of course two issues: 1. Inline functions are weak symbols and don't violate ODR 2. Static variables in C++ do violate ODR (btw in C they do not) so basically you COPY every instance of the functions across compilation unit. If you define some static function that does something by macro I'd rather expect to see stuff like inline void foo(...) { static foo_instance ... f = ...; f(...) } I don't know if it is implementable but it makes more scene or at least justify use of macro. On the other hand I don't see (with exception of some corner cases) why this is better than BOOST_FIT_STATIC_FUNCTION(foo) = fit:compose(bar1,bar2) // * calls bar1(bar2(...)) inline void foo(...) { bar1(bar2(...)) } or inlined auto foo = fit:compose(bar1,bar2); Is better than auto foo = [](...) { bar1(bar2(...) ) }; I understand (after getting this reply http://article.gmane.org/gmane.comp.lib.boost.devel/265453) That there are some meta-programming cases that the library is really helpful. But this isn't something I see in library documentation or it isn't the major library purpose. And for trivial cases I think direct approach is better as it is just clearer and easier to understand for an average code maintainer. > * Implementation Reviewed barely... no comment > * Documentation This is another major issue it does not clear what library does and how it is useful. This alone would make a condition on the acceptance. > * Tests Didn't reviewed > * Usefulness It is probably the major questions that the library need to ask. What is the purpose of the library and what problem it solves. I didn't find a complete answer to this question. > - Did you attempt to use the library? If so: No > - How much effort did you put into your evaluation of the review? Several hours - mostly reading docs and code. I can't say I reviewed most of the code or most of the docs. I stopped before as I couldn't find the satisfying answers regarding what library does. I understand that if somebody would submit the library regarding astrophysics or bio-modeling I'd had hard time to understand what library does. However regarding FIT I don't think it is the case. So if you (review manager/library designer) feel that I attempted to review a library in a field I don't get - feel free to ignore my review. Artyom Beilis

7 8

[release] New libraries documentation + release notes
by Daniel James 20 Mar '16

20 Mar '16

Hi, I'm starting to compile the release notes, so please add the info for your library. Easy way to do so is to click the pencil icon at the top right of the code at: https://github.com/boostorg/website/blob/master/feed/history/boost_1_61_0.q… Or you can just email me if you don't like quickbook. Also, the documentation for the new libraries in this release is now building okay and I've added links to the release notes, so please can the relevant people take a look: http://www.boost.org/users/history/version_1_61_0.html thanks, Daniel

3 4

[1.61] Two weeks remaining for new libraries and breaking changes
by Vladimir Prus 19 Mar '16

19 Mar '16

I wanted to remind that the master branch is closing for new libraries and breaking changes on Feb 24, about two weeks from now. We'll be closed for major changes on March 2. That's not a lot of time, please make sure any big work is completed. Thanks, -- Vladimir Prus http://vladimirprus.com

10 38

[Spirit X3] move_to
by Mikhail Strelnikov 18 Mar '16

18 Mar '16

Hi, I'm using Boost 1.60 with Spirit X3 from develop branch (because it has this fix <http://boost.2283326.n4.nabble.com/Single-element-attributes-in-X3-quot-sti…> ). If I change the rule from this: auto a2 = rule<class i2, ast::a2>{} = a1 >> a2_v2; to this: auto a2 = rule<class i2, ast::a2>{} = a1 >> '(' >> a2_v2 >> ')'; it stops compiling (both gcc and clang) with error "no matching function for call to 'move_to'". Here is full source code: #include <boost/spirit/home/x3.hpp> #include <boost/spirit/home/x3/support/ast/variant.hpp> #include <boost/fusion/include/define_struct_inline.hpp> #include <string> #include <vector> namespace ast { BOOST_FUSION_DEFINE_STRUCT_INLINE ( a1, (std::string, v) ) BOOST_FUSION_DEFINE_STRUCT_INLINE ( a2, (a1, v1) (std::vector<double>, v2) ) BOOST_FUSION_DEFINE_STRUCT_INLINE ( a3, (std::vector<a2>, v1) ) } using namespace boost::spirit::x3; auto a0 = rule<class i0, std::string>{} = lexeme[+alpha]; auto a1 = rule<class i1, ast::a1>{} = a0; auto a2_v2 = rule<class a2_v2, std::vector<double>>{} = -(double_ % ','); auto a2 = rule<class i2, ast::a2>{} = a1 >> '(' >> a2_v2 >> ')'; //auto a2 = rule<class i2, ast::a2>{} = a1 >> a2_v2; auto a3_v1 = rule<class a3_v1, std::vector<ast::a2>>{} = *(a2 >> ';'); auto a3 = rule<class i3, ast::a3>{} = a3_v1; using data = ast::a3; int main() { std::string script = ""; auto begin = script.begin(); auto end = script.end(); data v; bool ok = (boost::spirit::x3::phrase_parse(begin, end, a3, space, v) && begin == end); } Here is how it is compiled: #!/bin/bash wget --no-clobber http://netcologne.dl.sourceforge.net/project/boost/boost/1.60.0/boost_1_60_… tar -zxvf boost_1_60_0.tar.gz wget --no-clobber https://github.com/boostorg/spirit/archive/develop.zip unzip develop.zip mv ./boost_1_60_0/boost/spirit ./boost_1_60_0/boost/spirit_old g++ -std=c++14 test.cpp -I./spirit-develop/include/ -I./boost_1_60_0/ -o test

1 0

[Root Pointer] Benchmark
by Phil Bouchard 17 Mar '16

17 Mar '16

So I renamed block_ptr<> to root_ptr<> and the code is accessible from here: https://github.com/philippeb8/root_ptr I am sorry but I made a mistake in the benchmark again. root_ptr<> wasn't using the fast allocator. Now it does and the new benchmark is: unique_ptr (new): 42.3736 unique_ptr (make_unique): 59.6104 shared_ptr (new): 69.9175 shared_ptr (make_shared): 46.9994 shared_ptr (allocate_shared_noinit): 31.0648 root_ptr (new): 30.5549 Regards, -Phil

5 36

[ann][offtopic] GSoC with the STE||AR-Group
by Thomas Heller 17 Mar '16

17 Mar '16

Hi list, We are very sad that Boost didn't make it into GSoC this year. However, for students with enthusiasm for C++, Parallelism and Concurrency and looking for an interesting project for this summer, there are other opportunities. The STE||AR-Group (https://summerofcode.withgoogle.com/organizations/5183170693562368/) is looking for prospective students, we have a large variety of projects surrounding the HPX Parallel Runtime System (one of the reference implementations of the upcoming additions to the C++17 standard for parallelism and concurrency). You can find our project ideas list here: https://github.com/STEllAR-GROUP/hpx/wiki/GSoC-2016-Project-Ideas In addition, if you have a project idea you'd like to mentor that would fit into the scope of our organisation, we would be more than happy to welcome you! We are looking forward to your responses at hpx-users(a)stellar.cct.lsu.edu or on our IRC channel #ste||ar on Freenode! Best regards, Thomas -- Thomas Heller Friedrich-Alexander-Universität Erlangen-Nürnberg Department Informatik - Lehrstuhl Rechnerarchitektur Martensstr. 3 91058 Erlangen Tel.: 09131/85-27018 Fax: 09131/85-27912 Email: thomas.heller(a)cs.fau.de

1 0

[Fit] About supported C++ version
by Vicente J. Botet Escriba 16 Mar '16

16 Mar '16

Hi, I was wondering if Boost.Fit shouldn't be a library as Boost.Hana, a library that makes use of the last C++ version and avoids as much portability issues as possible. Which C++14 features will make the library simpler? Which adopted C++17 features will make the library simpler? Which other C++ features would be missing? Paul I guess that you are using a C++11 compiler and want to support this version, could you confirm? Best, Vicente

2 3

Providing means to verify integrity and authenticity for releases
by Daniel Hofmann 16 Mar '16

16 Mar '16

The current download page at > http://www.boost.org/users/download/ redirects the user to SourceForge for downloading sources and / or binary Boost distributions. SourceForge can no longer be trusted as a hosting platform, as you can for example see following this thread > http://lists.boost.org/boost-users/2016/02/85662.php where a user was tricked into downloading some arbitrary binary while downloading a Boost release. Unfortunately there does not seem to be a secure and convenient way to download Boost releases. Although Github's Boost "releases" can be found at > https://github.com/boostorg/boost/releases but those are only repository snapshots, from which you can not even build a Boost distribution. And whereas the Boost 1.60 rc1 announcement mail at least provides checksums > http://lists.boost.org/boost-users/2015/12/85435.php The official 1.60 release announcement mail does not > http://lists.boost.org/boost-users/2015/12/85467.php Correct me if I'm wrong, but there is no way for obtaining a Boost release and verifying its integrity and authenticity. The only option I'm seeing is recursively cloning all Boost repositories from Github and building a release by myself. Can we please change this situation? Here are some options that come to mind ordered by amount of effort: - Providing checksums - Educating users on the Downloads page - Signing releases with a trusted Release Team key - Changing the hosting platform Cheers, Daniel J H

7 8

[ann] C++ Archive Network
by Egor Pugin 16 Mar '16

16 Mar '16

Hi, I'd like to present my project - C++ Archive Network (or simply CPPAN). You can consider it as (yet another) C/C++ package manager. In the beginning it was targeting C++ Modules, but it seems it can handle many current C/C++ projects. You can read more information about idea and motivation from links below (see docs). This is an alpha release (0.1.2). I'm announcing it only here to get initial feedback (but still boost community is large enough). As always the general idea is to create programs faster, to reuse and share your or anyone's code. But specifically here: source only packages (no tools, no tests in distribution pkg - keep all your stuff in your repos), Modules-like paths (org.boost.algorithm), semantic versioning (semver.org) zero-conf. projects, all-or-nothing dependencies (no optional deps.) and some other convenient features. At the moment CPPAN works only with CMake. It generates files that you include to your project (see docs). To be closer to boost and modern C++ community I added some popular projects as demo versions: - boost 1.60.0 - all projects - some core parts work (filesystem, thread, log, asio etc.), did not tried less used libraries. - new boost libs: coroutine2 (added but not set up, so does not works), dll - boost incubator libs: fit, DI - you can try them faster during reviews - gsl, range-v3, mach7, sqlpp11 (and connectors) - some modern libs - concepts libs: cmcstl2, text_view - so users can try and adopt new C++ features faster - google's: gtest,gmock,protobuf, recent 'highwayhash' lib - sqlite3 (3.11.0) - zlib, bzip2 - libressl - image libs: libpng, tiff, gif, webp, some others - libpqxx Most of them just works. Almost all dependencies are met. In boost.* I do not untangle the core dependencies, so base dependency-ball is quite big. I believe that library authors can eliminate unneeded deps better. Links: https://cppan.org/ - website https://github.com/cppan/cppan - client, use issues for feedback https://github.com/cppan/cppan/wiki - docs https://github.com/cppan/demo_project - example project that uses cppan https://github.com/cppan/packages - example packages (config files) https://github.com/cppan/packages/blob/master/boost/boost.cppan - base boost config file (big list of deps, probably can be shorter) --- ps. I'm not writing about pros. and cons. Both are present. Just give CPPAN a try and we'll see how we can improve it. -- Egor Pugin

5 5

Re: [boost] [Block Pointer] Benchmarks
by Glen Fernandes 15 Mar '16

15 Mar '16

On Sun, Mar 13, 2016 at 1:02 PM, Phil Bouchard wrote: > Thanks for your help. > > So block_ptr<> wouldn't be a drop-in replacement for shared_ptr<> but an > add-on for: > - complex containers, > - neural networks, > - C# / Java / Javascript engines, > - etc. > No problem. Also renaming the thread to reflect the current benchmarks (since it doesn't appear that block_ptr is faster than shared_ptr where creation is involved). 1. Gathering these benchmarks is always useful, even if block_ptr was not designed to be a drop-in replacement for share_ptr. So you had the right idea, even if the execution needed some work. It's important to get the execution correct, though, otherwise the results are not meaningful - which is where I (and I believe Rob, also) was trying to steer you. 2. Looking at the numbers you obtained after modifying my benchmark example, they are close enough that at least you could say "Using block_ptr over shared_ptr won't be a significant performance loss": > unique_ptr (new): 47.7686 > unique_ptr (make_unique): 46.8545 > shared_ptr (new): 77.8261 > shared_ptr (make_shared): 50.8072 > shared_ptr (allocate_shared_noinit): 33.021 > block_ptr (new): 69.6554 3. There are more benchmarks beyond just creation (though creation is likely the most meaningfully expensive one). Copying overhead might be one: I haven't looked at what is involved in copying a block_ptr, but some work happens when you copy a shared_ptr. 4. shared_ptr by way of allocate_shared allows creation to not involve 'new' expressions, or even a call to '::operator new(std::size_t)' at all. i.e. For those C++ projects that have a requirement for all dynamic allocation in their project to involve some stateful custom allocator instances of some stateful custom allocator type, they can still use shared_ptr (with allocate_shared). Is this possible with block_ptr? I hope to take a look at the block_ptr motivation, design, and implementation when I have some time next week. The title of the thread caught my attention; "X is 600% faster than Y" always has a high excitement potential. Glen

8 16