May 2018 - Boost - lists.preview.boost.org

Enabling issues on boostorg/mpi
by Alain Miniussi 21 May '18

21 May '18

Hi, Could somebody with the sufficient permissions enable issues on the github mpi repository ? Thanks Alain Miniussi

2 2

Proposal for a thread synchronisation primitive: future_queue
by Martin Hierholzer 21 May '18

21 May '18

Dear boost developers, I want to propose a thread synchronisation primitive, which works like a mixture of (lockfree) queues and futures. Therefore I gave it the name "future_queue". It is basically a lockfree queue which can be waited on for new values (without busy waiting), but it has some nice extra features like when_any() and continuations. The github project can be found here: https://github.com/mhier/future_queue It already contains a README file with a short description, an example and results of a first performance measurement. Further documentation so far exists only inside the code: https://github.com/mhier/future_queue/blob/master/include/future_queue.hpp My first question is, whether this is really something new (I might have overlooked something in the many boost libraries) and if it makes sense to integrate it into the boost libraries. Another question would be whether this should really be its own library, since it is basically a single class (with a few helper classes) and a single non-member function. Maybe it would be better to make it part of an existing library? I am happy to receive some first feedback! Cheers, Martin -- Martin Hierholzer DESY Notkestrasse 85 22607 Hamburg Office 55a / 123 Tel: +49 40 8998 1897 Mob: +49 176 992 46 476 Skype: mhier48 This e-mail may contain confidential and/or privileged information and is intended only for the use of the individual or entity named above. If you received it in error, please advise the sender immediately by reply e-mail and delete the original. Any further use - especially any form of publication - of this e-mail without explicit authorisation is strictly prohibited.

4 11

[Hana] "basic_tuple.hpp" bug when building with VS2017
by Daniel Chen 20 May '18

20 May '18

Hello, Just installed Boost and Visual Studio 2017 a few days ago. Here is a bug when building with Hana's "basic_tuple.hpp" header with Visual Studio 2017: In basic_tuple.hpp: template <> struct drop_front_impl<basic_tuple_tag> { template <std::size_t N, typename Xs, std::size_t ...i> static constexpr auto drop_front_helper(Xs&& xs, std::index_sequence<i...>) { return hana::make_basic_tuple( detail::ebo_get<detail::bti<i+N>>(static_cast<Xs&&>(xs))... ); } template <typename Xs, typename N> static constexpr auto apply(Xs&& xs, N const&) { constexpr std::size_t len = detail::decay<Xs>::type::size_; return drop_front_helper<N::value>(static_cast<Xs&&>(xs), std::make_index_sequence< N::value < len ? len - N::value : 0 >{}); //<Bug here - line 197 of basic_tuple.hpp } }; Upon building with "basic_tuple.hpp", the line highlighted in green reports an error: "syntax error: unexpected token ')', expected 'id-expression'". Focus on the line above that one: N::value < len ? len - N::value : 0 If you add parentheses, the error goes away: (N::value < len) ? len - N::value : 0 A similar bug occurs - and is fixed likewise - in the "string.hpp" header file at line 275, that is: template <> struct drop_front_impl<string_tag> { ... template <char ...xs, typename N> static constexpr auto apply(string<xs...> const& s, N const&) { return helper<N::value>(s, std::make_index_sequence< N::value < sizeof...(xs) ? sizeof...(xs) - N::value : 0 >{}); //<bug here - line 275 of string.hpp } I'm not sure if this is a Microsoft compiler issue (I will post to them as well), but I feel like someone should add those parentheses (and then some) to the line since it is safer - the "<" can be easily confused for a template bracket instead of a less-than sign by a parser. Daniel Chen Infinera

2 1

Boost.Test, cygwin, and ISO/POSIX
by James E. King, III 19 May '18

19 May '18

Hello, I have been adding some CI infrastructure to the CMT repositories and found I am able to reproduce a failure building boost libraries that use Boost.Test for their unit tests. If you build with "cxxstd=03" you can get to some undefined functions. See: https://github.com/boostorg/test/issues/144 It looks like one needs to define _POSIX_C_SOURCE=200112L or cxxstd-dialect=gnu in order for these functions to become visible. I was thinking of addressing this two ways: 1. I want to change boost build (bjam) to define _POSIX_C_SOURCE when building on cygwin to this value if it is not otherwise defined. I can't find the right place to add this. 2. I don't think it's reasonable for Boost.Test to modify _POSIX_C_SOURCE since it could be included after sys/functional header, so I would like to either change Boost.Test so it does not depend on these three functions, or check to see if the right preprocessor definitions are set on cygwin and put up a better error message, or simply change the documentation and update all the CI build scripts in all the affected repositories to set _POSIX_C_SOURCE as part of the CI build. Any thoughts or other ideas on this? Thanks, Jim

1 0

disjoint_sets repo status
by James E. King, III 17 May '18

17 May '18

What is the current status of https://github.com/boostorg/disjoint_sets ? The headers appear to be in a "pending" state. There's almost nothing in this repo. Perhaps it was put here to solve a circular dependency issue? It looks like it is used by graph and graph_parallel. Could we move this into Boost.Algorithm? Thanks, Jim

1 0

[release][chrono] libs/chrono/stopwatches/include not in boost/chrono/stopwatches
by Andrey Semashev 17 May '18

17 May '18

Hi, In the 1.67 released archive, there is libs/chrono/stopwatches/include directory that contains headers that look like they should be placed in boost/chrono/stopwatches. With the checkout from git, `b2 headers` create links boost/chrono/stopwatches and boost/chrono/stopwatches.hpp that refer to that directory. Looks like the packaging scripts need to be updated.

2 1

dynamic_bitset needs some attention
by James E. King, III 16 May '18

16 May '18

I have been adding a common CI infrastructure to the CMT repositories. Occasionally I have to disable one or two build jobs to get a clean build and I file an issue. When I got to dynamic_bitset however, there's clearly something going on in there and it isn't good. I think it has to do with streaming. It's probably just one issue causing most of the failures. You can see the results from my latest dry-run PR here: https://github.com/jeking3/dynamic_bitset/pull/1 Right now both the main travis and all of the appveyor builds are failing, and they all look like they get stuck on test 4 in iostream processing. Something just isn't right in there - it could be the test itself or it could be something else. I may submit a PR into the repo for the CI additions anyway even if they are failing, since it points out this component has rotted and needs attention. Also if my PRs in pool and function can be merged, we'll have better CI on those. Aiming to get all the CMT repos covered with appveyor, travis, codecov, and eventually covscan. - Jim

1 0

Clarifying Boost evolution plans mentioned on Twitter
by Bryce Adelstein Lelbach aka wash 16 May '18

16 May '18

Hi all, Some pictures from the C++Now "Boost Library In A Week" end-of-week summary were posted to Twitter today by an attendee, along with some mistaken statements suggesting that "Boost plans to move to C++11". I want to clarify what is actually going on: 0.) During the week, the Library In A Week group discussed some ideas for how Boost might migrate to newer standards. I'm somewhat to blame as I initiated these discussions. 1.) Those ideas were collected into a draft document by Jeff Garland. 2.) During the closing panel, Jeff presented the draft document and stated the next step would be to share it with the mailing list. 3.) Due to an innocent misunderstanding, an attendee posted a picture of part of the document with the unclear message "Boost plans to move to C++11". 4.) I wrote this email. I have asked the attendee to delete the tweet (and repost with a clearer message if they wish). I don't want to share the document from LIAW. Jeff put a lot of work into it and I don't want to steal his thunder (I realize this email inevitably does that). I'm sure he will follow up with a post explaining what ideas we came up with soon. I'm very sorry for the confusion, and I'm sure the attendee is as well. This was just an innocent mixup. ------------------------------------------------------ Bryce Adelstein Lelbach aka wash ISO C++ Committee Member CppCon and C++Now Program Chair Thrust Maintainer, HPX Developer CUDA Convert and Reformed AVX Junkie Sleep is for the weak ------------------------------------------------------

4 3

[interprocess]Request to support wchar_t* file name on Windows
by 孙朝阳 15 May '18

15 May '18

Hi, I'm using boost.interprocess.file_mapping on Windows, and I need to use it to open files with localized file path, so I want file_mapping can support wchar_t file name on Windows. Does any reason prevent file_mapping to support wchar_t file name? If no, would someone like to add wchar_t supporting for file_mapping? I also created a pull request for this: https://github.com/boostorg/interprocess/pull/28 Thanks, Jackson Sun

1 0

boost libs compiled for gcc toolset with boost 1.67 and CMake 3.11 won't find libs
by Demetri Kopsas 12 May '18

12 May '18

Hi, I discovered that the latest version of boost does not work with CMake when using the GCC toolset. I submitted this bug: https://gitlab.kitware.com/cmake/cmake/issues/17974#note_408614 Using MSVC as the toolset is fine on windows, but using mingw (gcc compiler) cannot find the libs unless I take out ‘x32’ or ‘x64’ from the name after compilation.

2 1