- Boost - lists.preview.boost.org

[container] [1.59.0b1] Behavioral (runtime) breakage with flat_xxx + ordered_unique_range + non-default comparator
by Adam Merz 21 Jul '15

21 Jul '15

I've encountered a serious behavioral breakage for code using flat_xxx in conjunction with ordered_unique_range and a non-default comparator in 1.59.0b1. Here is a small repro: #include <initializer_list> #include <algorithm> #include <functional> #include <iterator> #include <iostream> #include <boost/container/flat_set.hpp> int main() { namespace bcont = boost::container; auto const p = [](auto const& desc, auto const& s) { std::cout << desc << ": "; std::copy( s.cbegin(), s.cend(), std::ostream_iterator<int>(std::cout, " ") ); std::cout << '\n'; }; auto const r = {4, 3, 2}; bcont::flat_set<int, std::greater<int>> const s{3, 2, 1}; p("orig set", s); auto s1{s}; s1.insert(r.begin(), r.end()); p("without\tordered_unique_range (expected output)", s1); auto s2{s}; s2.insert(bcont::ordered_unique_range, r.begin(), r.end()); p("with\tordered_unique_range (erroneous output)", s2); } I've worked around this locally by rerouting the ordered_unique_range codepath to the default codepath, but this seems like a showstopper for 1.59.0 as the code's behavior has broken entirely silently. Regards, Adam Merz

2 1

Python bindings for Boost.Interprocess
by Edison Gustavo Muenz 21 Jul '15

21 Jul '15

Hi all, this is my first e-mail on the list, so please be nice, I’m new here. I have this problem in which I need to communicate between two processes, one written in C++ (it’s actually a .so) and one written in python. They need to share large arrays. I’ve searched on the internet for a way to use shared-memory in python, and the only alternative I found was POSIX IPC for Python <http://semanchuk.com/philip/posix_ipc/> (which looks good, but it is *nix only). Also, I find the Boost.Interprocess library to be a joy to use, it worked for the problem I had, so I wanted to use it. Since I didn’t find python bindings for Boost.Interprocess is it desirable to have this? I started writing some code which uses a tiny part of Boost.Interprocess that is necessary to solve my problem. It is working already. I’m asking here to know what do boost experts think about this idea, if it unreasonable or if other developers already had this need. Feedback would be greatly appreciated. Thanks, Edison Gustavo Muenz

2 1

[mpl] Updated fixes on 'master'
by Edward Diener 18 Jul '15

18 Jul '15

A number of fixes in mpl, by Bruno Dutra and Deniz Bahadir, had been tested on 'develop' for quite a long time but had not been merged to 'master' for the upcoming release. The regression tests on 'develop' had shown that the fixes were working properly and I have taken the responsibility of merging them now to 'master' for the upcoming release. I will be checking the 'master' regression tests. If there are any problems I will revert this merge but I am 99.99% certain that the changes are correct, so I take full responsibility for merging. The fixes come from Trac items so I feel that correcting these bugs is important for the upcoming release else I would not have performed the merge and the push to Boost.

1 0

[string_ref] [utility] Add string_view?
by Beman Dawes 18 Jul '15

18 Jul '15

Boost has shipped string_ref since 1.53. See www.boost.org/doc/libs/1_58_0/libs/utility/doc/html/string_ref.html In the meantime, the C++ committee has shipped a final version, named string_view, in the Library Fundamentals TS. See http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4480.html#string.v… In addition to the name change, the interface in the TS more closely mimics the std::basic_string interface. I'd really like to see the final TS basic_string_view added to Boost so I can start using it now. Using basic_string_ref, and then later having to rename all uses, is less attractive. I suspect others may feel the same way. Because a Boost basic_string_view would be an implementation of a standard library TS class, and because basic_string_ref is already in Boost, it seems a min-review would be sufficient rather than a formal review. So what I'm proposing here is to create a basic_string_view branch of the utility repo, and start working on basic_string_view there. Once it is reasonably compliant with the TS and Boost doc requirements we would hold a mini-review and then merge the branch to develop, and once stable, to release. Comments? --Beman

6 8

Boost 1.57.0 WinCE support/maintenance
by Dennis Driesens 17 Jul '15

17 Jul '15

Hello, We're porting a software product, which uses boost 1.57.0 to Windows embedded compact 7 . In this context I have a few questions. What is the support /maintenance status of boost for Windows embedded compact 7? We patched boost 1.57.0, to make it build for WEC7. Is it possible to have these patches included in boost? Regards, Dennis Driesens Dennis Driesens * Sr. Software Engineer * PU Custom Systems * Dennis.Driesens(a)tomtom.com<mailto:Dennis.Driesens@tomtom.com> * +31 (0) 40 84 44793 office * www.tomtom.com<http://www.tomtom.com>

2 1

svn.boost.org wiki
by Stefan Seefeld 17 Jul '15

17 Jul '15

Hello, I have sent the following to the Boost documentation list, but since this is of broader relevance I have been asked to resend it here... The content of https://svn.boost.org/trac/boost/ seems rather outdated, and it has been confusing me a number of times as it wasn't clear what is and what isn't up-to-date. * There is an entire section "Quick Access to the Boost Subversion Repository". Shouldn't that be removed entirely ? * The section "Git and Modular Boost" starts with "Boost is moving...". Isn't that conversion complete (and thus, shouldn't the docs state that clearly) ? * There are other seemingly outdated bits, but there are relatively minor... What I was actually trying to figure out is how / where to indicate that for Boost.Python I would like issues to be submitted with the respective github issue tracker, rather than the old Boost Trac instance. It's still not clear how to do that (as the Boost docs are again quite monolithic in this respect, i.e. don't allow a per-library policy). So if anyone has suggestions on where / how to express that, please let me know. Thanks ! Stefan -- ...ich hab' noch einen Koffer in Berlin...

4 6

Master branch is open again - for bug fixes, doc updates, etc
by Marshall Clow 16 Jul '15

16 Jul '15

Now that we've shipped the 1.59.0 beta, the master branch is open again for updates. Please do not add new features, etc to the master until after the release. But bug fixes, documentation updates, etc are very welcome. Note that: * GCC 5.2 was released today * VS2015 should be out next week * Clang 3.7 branched for release this week Compatibility with these new compiler releases would be a good thing :-) -- Marshall

1 0

[Iterator][Type_Erasure] Conflicting declaration. (Fix is provided.)
by Deniz Bahadir 16 Jul '15

16 Jul '15

Hi there, I noticed a "conflicting declaration" error when developing a program which uses Boost.Program_Options and Boost.Type_Erasure. Further investigation showed me, that the conflict is actually between Boost.Iterator and Boost.Type_Erasure. A test-case which demonstrates this compiler-error is the following simple two-liner: #include <boost/type_erasure/iterator.hpp> #include <boost/iterator/iterator_adaptor.hpp> Clang is quite verbose and outputs the following: /tmp/boost-1_58/boost/iterator/iterator_adaptor.hpp:44:18: error: target of using declaration conflicts with declaration already in scope using iterators::use_default; ^ /tmp/boost-1_58/boost/iterator/iterator_adaptor.hpp:40:10: note: target of using declaration struct use_default; ^ /tmp/boost-1_58/boost/type_erasure/iterator.hpp:27:8: note: conflicting declaration struct use_default; ^ 1 error generated. The fix is actually quite simple and should be applied to Boost.Iterator. I already provided a pull-request [1] for this. I hope this fix can still make it into the upcoming 1.59 release of Boost. Best regards, Deniz Bahadir [1] https://github.com/boostorg/iterator/pull/15 -- BENOCS GmbH Dipl.-Inform. Deniz Bahadir Winterfeldtstr. 21 10781 Berlin Germany Phone: +49 - 30 / 577 0004-22 Email: deniz.bahadir(a)benocs.com www.benocs.com Board of Management: Michael Wolz, Dr.-Ing. Oliver Holschke, Dr.-Ing. Ingmar Poese Commercial Register: Amtsgericht Bonn HRB 19378

3 2

Boost.DLL formal review is ongoing
by Vladimir Prus 16 Jul '15

16 Jul '15

The formal review of Boost.DLL library by Antony Polukhin is nearing the end of the first week. As of today, over 100 people read through the library's tutorial, suggesting quite some interest. However, no reviews were submitted yet. Please take the time to post your thoughts - even if you don't have time to do a full review, or try the library, comments on design and interfaces are still very valuable. The summary of the library features and review checklists are reproduced below. Boost.DLL is a C++98 library for comfortable work with DLL and DSO. Library provides a portable across platforms way to: - load libraries at runtime - import and export any native functions and variables - make alias names for C++ mangled functions and symbols - query libraries/objects and executables for sections and exported symbols - self loading and self querying - getting program and module location by exported symbol The documentation can be found at: http://apolukhin.github.io/Boost.DLL/index.html and the source can be obtained at: https://github.com/apolukhin/Boost.DLL Please post your reviews on the mailing list and if possible, answer the following questions: - Should the library be accepted? - How useful is it? - What's your evaluation of - Design - Implementation - Documentations - Tests - How much effort did you put into your evaluation? - Did you attempt to use the library? On what systems and compilers? Thanks, Volodya _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

16 42

[afio] AFIO review postponed till Monday
by Niall Douglas 16 Jul '15

16 Jul '15

Just a quick heads up that the AFIO review is postponed till Monday to give me more time to refactor the docs to match the final API. Some quick notes on what will be presented for review, so there are no surprises: 1. I won't have the time to rewrite the code examples into C++ 1z coroutines before the review unfortunately, so you'll get a warning sign in the docs saying that is coming. The code example this matters most to is the find regex in files example as the code is very messy (full of callback handlers) without coroutines. 2. The API presented is expected to be a final API, but the implementation will be a messy conglomeration of the existing mature dispatch engine with the new lightweight future-promises. In other words, code will compile and run, but the internal engine is going to be completely replaced eventually as I debug lightweight future-promises. As lightweight future-promises is all brand new code mostly written between 12am and 4am each night since May after a work day, I don't hugely trust it enough to present you guys with an engine based on it. Just because it passes a conformance suite doesn't mean it is without quirks - plus, I discovered some surprises in the Concurrency TS about which I'll need to get some clarifications. As I mentioned though, the user facing API will be final, as promised. And it does use monads and lightweight future-promise. 3. Minimum compilers will unfortunately be GCC 4.8, clang 3.2 and VS2015. Sorry about that, lightweight monadic future-promises substantially bumped the compiler requirements, even as it is I have several GCC 4.8 specific workarounds. Niall -- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/

1 0