- Boost - lists.preview.boost.org

Re: [boost] [Boost-users] [process] Formal review results
by Bjorn Reese 13 Nov '16

13 Nov '16

On 11/09/2016 07:45 PM, Antony Polukhin wrote: > The Boost.Process library is accepted. Unconditionally? Please notice that some of the votes were conditional, and if those conditions are not met, they should be counted as no votes. > The most controversial part of the library is it's interface. Many > comments were addressing: > * passing arguments > * starting process right in `system(...)` function The comments were about starting the process in the child constructor.

4 3

Boost.Config for clang on Linux with libc++
by Belcourt, Kenneth 13 Nov '16

13 Nov '16

Hi, Attached are user.hpp and the console output from running configure with clang-3.9. Note that I tried to configure using libc++ but the configure script immediately failed, so these are the results for clang-3.9 with the default stdlib (libstdc++). Can someone please take a look and see what adjustments might need to be made to Boost.Config to support this configuration? Once it works for the default stdlib, I’d like to rerun it to use libc++. Thanks, Noel Belcourt

3 4

[process] Formal Review
by Nat Goodspeed 13 Nov '16

13 Nov '16

I would like to surface a number of issues, large and small. You are clearly proud to support a number of different syntactic ways to express the same semantic operation. To me this is a minor negative. While in theory it sounds nice to be able to write Process consumer code any way I want, in a production environment I spend more time reading and maintaining other people's code than I do writing brand-new code. Since each person contributing code to our (large) code base will select his/her own preferred Process style, I will be repeatedly confused as I encounter each new usage. I admire the support for synchronous pipe I/O, while remaining skeptical of its practical utility. Synchronous I/O with a child process presents many legitimate opportunities for deadlock: traps for the unwary. I would be content with a combination of: * async I/O on pipes (yes, using Asio) * system() for truly simple calls * something analogous to Python's subprocess.Popen.communicate(): pass an optional string for the child's stdin, retrieve a string (or, say, a stringstream) for each of its stdout and stderr. The example under I/O pipes the stdout from 'nm' to the stdin of 'c++filt'. But the example code seems completely unaware that c++filt could be delayed for any of a number of reasons. It assumes that as soon as nm terminates, c++filt must have produced all the output it ever will. I worry about the Process implementation being confused about such issues. I'm dubious about the native_environment / environment dichotomy. As others have questioned, why isn't 'environment' a typedef for a map<string, string> (or unordered_map)? I understand that you desire to avoid copying the native process environment into such a map until necessary, but to me that suggests something like an environment_view (analogous to string_view) that can perform read-only operations on either back-end implementation. Operations involving splitting and joining on ':' or ';' should be defined purely in terms of strings and ranges of strings. They should not be conflated with environment-map support. The documentation so consistently uses literal ';' as the PATH separator that I worry the code won't correctly process standard Posix PATH strings. At this moment in history, an example showing integration of Process with Boost.Fiber seems as important as examples involving coroutines. Why is the native_handle_t typedef in the boost::this_process namespace? While in full context it makes sense to speak of "assigning" an individual process to a process group, the method name assign() has conventional connotations. Use add() instead. There's a Note that says: "If a default-constructed group is used before being used in a process launch, the behaviour is undefined." I assume you mean "destroyed before being used," but this is a concern. If a program has already instantiated a process group, but for any reason decides not to (or fails to) launch any more child processes, does that put the entire parent process at risk? What remedial action can it take? Move the group object to the heap somewhere and let it leak? Spawn a bogus child process for the purpose of defusing the ticking process group instance? If you're going to reify process group at all, you should wrap more logic around it to give it well-defined cross-platform behavior. And it should definitely be legal to create and destroy one without associating any child process with it. Given support elsewhere for splitting/joining PATH strings, the string_type path parameter to search_path() feels oddly low-level. Maybe accept a range of strings? > At a minimum, kindly state: > - Whether you believe the library should be accepted into Boost > * Conditions for acceptance YES, IF the present Boost.Process preserves the ability of its predecessor to extend it without modifying it. I'm sorry, thorough reading of the documentation plus some browsing through the code leaves me doubtful. Examples of such extensions: * With Boost.Process 0.5 I can use Boost.TypeErasure to pass an "initializer" which is a runtime collection of other initializers. Such a feature in Process 0.6 would support people's requests to be able to assemble an arbitrary configuration with conditional elements and use it repeatedly. * I can create an initializer (actually one for each of Posix and Windows, presenting the same API, so portable code can use either transparently) to implicitly forcibly terminate the child process being launched when the parent process eventually terminates in whatever way. Please understand that I am not asking for the above features to be absorbed into the Process library: I am asking for a Process library extensible enough that I can implement such things with consumer code, without having to modify the library. Perhaps Process 0.6 already is! It's just hard for me tell. Another feature should, in my opinion, be incorporated into the library: * On Windows, if you desire to pass any file handles at all to the new child process, it is completely whimsical what *other* open file handles you may inadvertently pass -- unless you play games with STARTUPINFOEXA and PROC_THREAD_ATTRIBUTE_LIST to enumerate exactly the set of handles you intend to pass. If the library doesn't natively support that, I *must* be able to pass a custom initializer with deep access to the relevant control blocks to set it up. This is a showstopper, as in "you can't remove that file because, unbeknownst to you, some other process has it open." > - Your knowledge of the problem domain I have hand-written process-management code in C++ several times before -- each with multiple implementations to support multiple platforms. I have tested the previous candidate Boost.Process 0.5 with a number of programs exercising a number of features. > You are strongly encouraged to also provide additional information: > - What is your evaluation of the library's: > * Design Design notes are at the top of this mail. > * Implementation I have only glanced over parts of the implementation. It seems somewhat more obscure than the corresponding pieces of Process 0.5, which is why I couldn't quickly satisfy myself as to the library's extensibility. > * Documentation Others have noted that the documentation is very sketchy in places. I too wish for more explanation. Much of the time I spent on this review was reading through the documentation. Apologies if I have misunderstood the library's capabilities. > * Usefulness I have felt for years now that it is essential to get a child-process management library into Boost. It grieves me to have to write platform-specific API calls into different applications. > - Did you attempt to use the library? I did not, sorry, ran out of time -- as you can infer from this review arriving at the end of the final day! > - How much effort did you put into your evaluation of the review? Most of my time was spent reading the documentation. I looked through a couple of the implementation files for further information.

4 8

Visual Studio 2015 Update 3 has removed std::unary_function and std::binary_function
by Marshall Clow 12 Nov '16

12 Nov '16

... if you build with /std:c++latest. See https://svn.boost.org/trac/boost/ticket/12586 for an example. Other vendors will be removing these types in C++17 mode in the future (or at least requiring a -D RETAIN_REMOVED_STUFF=1 or something like that) A quick grep for these terms found 153 instances of "std::unary_function" and 118 of "std::binary_function" across several libraries, including accumulators, algorithm, config, container, function, gil, graph, icl, mpi, msm, polygon, ptr_container, serialization, smart_ptr, tr1, unordered, utility - and probably others. I think that we ought to have a boost-wide solution for this (i.e, everyone should solve this the same way). I can think of three possibilities: * Deny, deny, deny: Boost doesn't work with C++17. Not my recommendation. * Since std::unary|binary_function is an empty struct with a set of two (or three) typedefs, just define the typedefs everywhere. Note that the typedefs (result_type, first_argument_type, second_argument_type) were deemed to not be useful in C++17 due to the existence of advanced type traits, etc. * Define a boost:unary_function and boost::binary_function struct that mimics the one that was in namespace std, and change over to use that. * Something else. Comments? -- Marshall

15 74

Adding any_dynamic_cast?
by Peter Dimov 12 Nov '16

12 Nov '16

I stumbled across https://github.com/bytemaster/Boost.DynamicAny which implements, in a fully backward compatible manner, dynamic cast functionality for boost::any. That is, it can give you X* from an 'any' holding a value of type Y if Y derives from X. It achieves this by simply changing template<typename ValueType> class holder : public placeholder { ValueType held; to template<typename ValueType> class holder : public placeholder, public ValueType which then enables dynamic_cast from holder<>* to OtherValueType*. This is pretty cool, and I don't remember seeing this idea anywhere else. Perhaps we should adopt it? One might even argue that this is how any_cast should always have worked, but I suppose we can't change it now for compatibility reasons. Although it's an interesting question whether C++17 'any' shouldn't be enhanced in this manner.

2 2

Request for a new submodule, tools/depinst
by Peter Dimov 12 Nov '16

12 Nov '16

I'd like to request the creation of a new submodule, tools/depinst, that will contain a Python script depinst.py which would, given a module in libs/, install the dependencies needed for running that module's tests. Currently, the .travis.yml files I use enumerate the dependencies manually, which is a bit tiresome: https://github.com/boostorg/function/blob/develop/.travis.yml - git submodule init libs/algorithm - git submodule init libs/align ... - git submodule init libs/type_traits - git submodule init libs/typeof - git submodule init libs/utility - git submodule init libs/winapi - git submodule init tools/build - git submodule update --depth 1 This will be replaced by just - git submodule update --init --depth 1 tools/depinst - git submodule update --init --depth 1 tools/build - python tools/depinst/depinst.py function

5 19

[CI] Activation for Boost.Units
by Jürgen Hunold 11 Nov '16

11 Nov '16

Hi Rene, could you please activate CI via Travis CI and Appveyor for Boost.Units? I hope that I have successfully adapted the scripts from Boost.Predef and that they work for Units as well. Is there any way to get myself some admin rights so I can scheduled cron jobs on TravisCI? I'd like to have at least a weekly run in order to check that no other library break units. And I noticed that the scripts don't work for my fork on GitHub. It needs to detect that and add a remote to the submodule in the Boost checkout somehow. Do you have any plans for that? Yours, Jürgen -- * Dipl.-Math. Jürgen Hunold ! * voice: ++49 4257 300 ! Fährstraße 1 * fax : ++49 4257 300 ! 31609 Balge/Sebbenhausen * jhunold(a)gmx.eu ! Germany

2 1

[Beast] std::to_string not available in gcc 4.8?
by Vinnie Falco 11 Nov '16

11 Nov '16

A user is complaining that std::to_string is not available in their distribution. Beast doesn't support toolchains that don't have full support for C++11. However, if this is the only thing missing maybe I can just use a Boost equivalent to get him up and running. I see boost::to_string in <boost/exception/to_string.hpp> is that a public interface? It doesn't look documented. What do folks typically do when they want to call std::to_string on an unsupported distribution?

4 4

libs/regex/build/Jamfile.v2 missing "import testing"
by Peter Dimov 10 Nov '16

10 Nov '16

In a partial installation (not all submodules installed), running b2 in the Boost root says: libs\regex\build\Jamfile.v2:93: in modules.load ERROR: rule "unit-test" unknown in module "Jamfile<D:\tmp2\boost-root\libs\regex\build>". because libs/regex/build/Jamfile.v2 uses the unit-test rule, defined in testing.jam, but does not have "import testing ;". Adding that line fixes it. I've no idea why it works in a full installation - probably because someone else imports testing first before it gets to regex.

2 1

[process] Formal review results
by Antony Polukhin 09 Nov '16

09 Nov '16

The Boost.Process library is accepted. Congratulations to Klemens, and thanks to everybody who reviewed the library! The votes were as follows: + Johannes - Niall Douglas +/- Tom Kent +/- Nat Goodspeed + Paul A. Bristow +/- Bjorn Reese + Edward Diener + Boris Schäling +/- Raphaël Londeix + Klaim - Joël Lamotte Legend: + accept - decline +/- accept conditionally Separate thanks to Gavin Lambert for providing a solution to deadlocking problem. Additional thanks to Nat Goodspeed, Niall Douglas, Gavin Lambert, Raphaël Londeix and Bjorn Reese for spending significant amount of time on discussing, reviewing or testing the library. -- Quick review summary -- During the review multiple technical issues were raised and as far as I know, Klemens already working on them. The biggest win of the review - is a fix for possible deadlock with pipes. The most controversial part of the library is it's interface. Many comments were addressing: * passing arguments * starting process right in `system(...)` function Current approaches have benefits and drawbacks, other approaches have different benefits and drawbacks. At this point some people agree that the current approach is the lesser evil. Reviewers also wished to have better/richer documentation. P.S.: Please continue with the discussion and report issues that would help to improve the library. -- Best regards, Antony Polukhin

3 3

[dll] trouble building boost dll with Intel compiler on Windows
by Blower, Melanie 09 Nov '16

09 Nov '16

Hello, We are having trouble building boost_dll version 1.61 with Intel compiler (version 17.0) on Windows with vs2015 (and other versions of visual studio). There's a problem because the macro configuration causes alias.hpp to use gnu-like syntax to specify section names: we don't support this syntax in our Windows compiler. Instead, we support "#pragma section" just like the Microsoft compiler. Alias.hpp would use the "correct" configuration for our Windows compiler "#if BOOST_COMP_MSVC" (reference the file boost/dll/alias.hpp) However, the file boost/predef/compiler/visualc.hpp sets BOOST_COMP_MSVC to BOOST_VERSION_NUMBER_ZERO which is 0 I propose that alias.hpp could use BOOST_COMP_MSVC_BUILD instead of BOOST_COMP_MSVC. Does that seem reasonable? Thanks and regards, Melanie Blower (I work for Intel on the Intel C++ compiler)

5 11

[fusion] trouble building boost fusion with Intel compiler on Windows and Linux
by Blower, Melanie 09 Nov '16

09 Nov '16

Hello, We are having trouble building boost_fusion version 1.61 with Intel compiler (version 17.0) on Linux, and on Windows with vs2015 (and other versions of visual studio). The problem can be fixed if I made this change to boost/fusion/container/vector/detail/config.hpp : // Sometimes, MSVC 12 shows compile error with std::size_t of template parameter . #if BOOST_WORKAROUND(BOOST_MSVC, BOOST_TESTED_AT(1800)) \ || BOOST_INTEL /* add this line */ # if defined(BOOST_FUSION_HAS_VARIADIC_VECTOR) # undef BOOST_FUSION_HAS_VARIADIC_VECTOR # endif #endif Symptom on Windows: ile ..\..\..\bin.v2\libs\fusion\test\fold.test\cc848a939769f1f7cf08091b8a228ddb\algorithm\fold.obj.rsp "algorithm\fold.cpp" -Fo"..\..\..\bin.v2\libs\fusion\test\fold.test\cc848a939769f1f7cf08091b8a228ddb\algorithm\fold.obj" -TP /Z7 /Od /Ob0 /W3 /GR /MDd /Zc:forScope /Zc:wchar_t /nologo /Qwn5 /Qwd985 -Qoption,c,--arg_dep_lookup /Qvc14 -Qpchi- /wd4675 /EHs -c -DBOOST_ALL_NO_LIB=1 "-I..\..\.." compile-c-c++ ..\..\..\bin.v2\libs\fusion\test\fold.test\cc848a939769f1f7cf08091b8a228ddb\algorithm\fold.obj icl @"..\..\..\bin.v2\libs\fusion\test\fold.test\cc848a939769f1f7cf08091b8a228ddb\algorithm\fold.obj.rsp" fold.cpp ../../../boost/fusion/container/vector/vector.hpp(71): error: namespace "boost::fusion::vector_detail::result_of" has no member "value_at_c" : boost::is_convertible<Sequence, typename result_of::value_at_c<This, 0>::type> ^ detected during: instantiation of class "boost::fusion::vector_detail::is_convertible_to_first<Sequence, This, <unnamed>> [with Sequence=int, This=boost::fusion::vector<int>, <unnamed>=1]" at line 305 instantiation of "boost::fusion::vector<T...>::vector(U &&...) [with T=<int>, U=<int>]" at line 159 of "algorithm\fold.cpp" Symptom on Linux: The original failure on Linux (zip.test in boost_fusion) "icpc" -c -xc++ -O0 -g -w1 -inline-level=0 -fPIC -m64 -DBOOST_ALL_NO_LIB=1 -I"../../.." -c -o "../../../bin.v2/libs/fusion/test/zip.test/intl-lnx/dbg/algorithm/zip.o" "algorithm/zip.cpp" ../../../boost/fusion/container/vector/vector.hpp(230): error: pack "U" does not have the same number of elements as "I" : store<I, T>(std::forward<U>(var))... ^ detected during: instantiation of "boost::fusion::vector_detail::vector_data<boost::fusion::detail::index_sequence<I...>, T...>::vector_data(boost::fusion::vector_detail::each_elem, U &&...) [with I=<0UL, 1UL>, T=<boost::fusion::vector_iterator<const boost::fusion::vector<int, int>, 0>, boost::fusion::vector_iterator<const boost::fusion::vector<char, char>, 0>>, U=<const boost::fusion::transform_view<const boost::fusion::vector<const boost::fusion::vector<int, int> &, const boost::fusion::vector<char, char> &>, boost::fusion::detail::poly_begin, boost::fusion::void_> &>]" at line 306 instantiation of "boost::fusion::vector<T...>::vector(U &&...) [with T=<boost::fusion::vector_iterator<const boost::fusion::vector<int, int>, 0>, boost::fusion::vector_iterator<const boost::fusion::vector<char, char>, 0>>, U=<const boost::fusion::transform_view<const boost::fusion::vector<const boost::fusion::vector<int, int> &, const boost::fusion::vector<char, char> &>, boost::fusion::detail::poly_begin, boost::fusion::void_> &>]" at line 41 of "../../../boost/fusion/view/zip_view/zip_view_iterator.hpp" Thanks and regards, Melanie Blower (I work for Intel on the Intel C++ compiler)

3 4

test: Initializiation of data test cases
by Florian Lindner 09 Nov '16

09 Nov '16

Hello, I would like to use data test cases. The problem that comes with that, is that I want to use objects that needs the application to be initialized before they are instantiated. Example is that an object that uses MPI needs to habe MPI_Init called before. This example code shows the order in which different hooks are run: #define BOOST_TEST_DYN_LINK // #define BOOST_TEST_MAIN #include <boost/test/unit_test.hpp> #include <boost/test/data/test_case.hpp> #include <boost/test/data/monomorphic.hpp> #include <mpi.h> #include <iostream> #include <string> #include <vector> using namespace std; struct GlobalFixture { GlobalFixture() { cout << "GlobalFixture" << endl; } ~GlobalFixture() { } }; BOOST_GLOBAL_FIXTURE(GlobalFixture); struct Test : public std::string { // derive from string to get operator<< that boost requires Test() { cout << "Object to test" << endl; } }; namespace bdata = boost::unit_test::data; template <class T> class RuntimeDataset { public: using sample = typename T::value_type;; enum { arity = 1 }; T data; explicit RuntimeDataset(T _data) { cout << "Dataset constructor" << endl; data = _data; } bdata::size_t size() const { return data.size(); } typename T::const_iterator begin() const { return data.cbegin(); } }; namespace boost { namespace unit_test { namespace data { namespace monomorphic { template <class T> struct is_dataset<RuntimeDataset<T>> : boost::mpl::true_ {}; }}}} struct Fixture { Fixture() { cout << "Fixture" << endl; } }; BOOST_AUTO_TEST_SUITE(MySuite) BOOST_DATA_TEST_CASE_F( Fixture, test1, RuntimeDataset<std::vector<Test>>( { Test() } ), // This should be evaluated when the test is about to run input) { BOOST_TEST(true); cout << "Data test case" << endl; } BOOST_AUTO_TEST_SUITE_END() // initialization function: bool init_unit_test() { cout << "init_unit_test" << endl; return true; } // entry point: int main(int argc, char* argv[]) { cout << "main" << endl; int retCode = boost::unit_test::unit_test_main( &init_unit_test, argc, argv ); return retCode; } I've removed some code parts, but it should compile right away using mpic++ -std=c++11 -lboost_unit_test_framework boosttesting.cpp Its output is Object to test Dataset constructor main init_unit_test GlobalFixture Running 1 test case... Fixture Data test case *** No errors detected So, before I have any chance to do my own initialization code, the Test object and the dataset is constructed. I have posted this to the user mailing list, but still looking for a way to work around that. Do you have any idea how I can do initialization before data is instantiated? Thanks, Florian

1 0

cross-compiling linux->osx: file was built for archive which is not the architecture being linked (x86_64)
by Frédéric Bron 09 Nov '16

09 Nov '16

I am trying to cross-compile for osx from linux using clang. When I build boost, I use the following: CC="x86_64-apple-darwin13-clang" CFLAGS="-O2 -DNDEBUG" CXX="x86_64-apple-darwin13-clang++-libc++" CXXFLAGS="-O2 -DNDEBUG -std=c++14" AR="x86_64-apple-darwin13-ar" AS="x86_64-apple-darwin13-as" RANLIB="x86_64-apple-darwin13-ranlib" ./bootstrap.sh --without-icu --prefix=/softs/usr/darwin13-clang/release/boost --with-toolset=clang Then I use the following user-config.jam: using clang : : x86_64-apple-darwin13-clang++-libc++ : <cxxflags>-std=c++14 ; with the following command: ./b2 -d2 -j8 --layout=system --user-config=user-config.jam --without-mpi --without-python --without-context --without-coroutine link=static threading=multi runtime-link=shared target-os=darwin threadapi=pthread address-model=64 define=BOOST_THREAD_USE_LIB define=BOOST_LOG_WITHOUT_EVENT_LOG variant=release install Apart from a few errors, everythings looks fine: ...failed updating 3 targets... ...skipped 3 targets... ...updated 13684 targets... However, when I link against the static libraries that have been built, I get this type of messages: ld: warning: ignoring file /softs/usr/darwin13-clang/release/boost/lib/libboost_system.a, file was built for archive which is not the architecture being linked (x86_64): /softs/usr/darwin13-clang/release/boost/lib/libboost_system.a I suspect that it is because the wrong ar and ranlib where used because I find in my log that b2 used "ar" and "ranlib" instead of "x86_64-apple-darwin13-ar" and "x86_64-apple-darwin13-ranlib". So I changed my user-config.jam to: using clang : : x86_64-apple-darwin13-clang++-libc++ : <cxxflags>-std=c++14 <archiver>x86_64-apple-darwin13-ar <ranlib>x86_64-apple-darwin13-ranlib ; but b2 is still using "ar" and "ranlib". Any idea of what my problem is? Frédéric

1 1

Support for fixed point arithmetic in Boost?
by Juan Dent 08 Nov '16

08 Nov '16

Hi, Sorry if this is a dumb question but I have been looking for fixed point types in Boost and not found any... Yet I feel there must be... Thanks for any help Juan

4 5

Fw: Build failed: smart_ptr 1.0.31-develop
by Peter Dimov 08 Nov '16

08 Nov '16

I'm getting a weird error from the CI: https://ci.appveyor.com/project/pdimov/smart-ptr/build/1.0.31-develop C:/projects/boost-root/tools/build/src\build-system.jam:982: in load *** argument error * rule boostcpp.post-build ( ok ? ) * called with: ( ok ok ) * extra argument ok C:/projects/boost-root\boostcpp.jam:513:see definition of rule 'boostcpp.post-build' being called C:\projects\boost-root\tools\build\src/kernel\modules.jam:295: in import C:\projects\boost-root\tools\build\src/kernel/bootstrap.jam:139: in boost-build C:\projects\boost-root\boost-build.jam:17: in module scope Command exited with code 1 This is Submodule path 'tools/build': checked out '32113ca3ddbc637d08e85c44502f3149d37323e0'

2 1

[process] The Formal review preiod is nearing the end
by Antony Polukhin 07 Nov '16

07 Nov '16

Dear Boost community, The formal review of Klemens David Morgenstern's Process library ends on 5th November. If you're willing to write a review - please hurry up! If you're willing to write the review but could not make it on time - please give a hint. In that case it is possible to postpone the final decision for a few days. Thanks in advance! -- Best regards, Antony Polukhin

5 8

[process] Formal review has ended
by Antony Polukhin 07 Nov '16

07 Nov '16

The library review has ended. I will need a few more days to calculate the votes and to wait for some people that spend a lot of time to make their vote more explicit. At this point, following people voted: * Johannes * Niall Douglas * Tom Kent * Paul A. Bristow * Bjorn Reese * Edward Diener * Boris Schäling If I've missed someone, please notify me. -- Best regards, Antony Polukhin

2 2

[iostreams] Last develop commit breaks release builds.
by Rene Rivera 07 Nov '16

07 Nov '16

FYI.. This commit: < https://github.com/boostorg/iostreams/commit/0a87de8932641a3896f49872a3c034… > Is causing release build errors. See the CI logs for details.. <https://circleci.com/gh/boostorg/boost/2190> Please address ASAP. -- -- Rene Rivera -- Grafik - Don't Assume Anything -- Robot Dreams - http://robot-dreams.net -- rrivera/acm.org (msn) - grafikrobot/aim,yahoo,skype,efnet,gmail

2 2

[config] std::shuffle support
by Edward Diener 07 Nov '16

07 Nov '16

Does Boost config currently have a macro to determine whether the c++11 std::shuffle algorithm is supported for a particular compiler ? If not supporting /std:c++latest by changing std::random_shuffle to std::shuffle will not be possible for c++03 libraries.

3 6

interest in structure of arrays container?
by Michael Marcin 06 Nov '16

06 Nov '16

Structure of arrays is a pretty well known pattern for improving memory density and enabling vectorization in high performance code. Also known as parallel arrays[1]. After hand-rolling yet another of these it occurred to me that you make a reasonable first implementation with a simple alias template. template< typename... Ts > using soa = std::tuple< std::vector<Ts>... >; example: soa<int,float,std::string> data; This is decent start but has several issues including: * nothing to preserve the invariant keeping the arrays the same length * missing coalesced interface for vector methods (size, reserve, etc) * no iterator * wasted storage for duplicated allocators, capacities, sizes * sizeof...(Ts) allocations could be a single large block The work required to address these issues is not too great I think. Potential names: soa<Ts> structure_of_arrays<Ts> parallel_array<Ts> struct_array<Ts> parallel_vector<Ts> struct_vector<Ts> Does this already exist somewhere? Does it seem useful? [1]https://en.wikipedia.org/wiki/Parallel_array

10 79

Re: [boost] [process] Formal Review starts today, 27 October
by Bjorn Reese 06 Nov '16

06 Nov '16

On 10/30/2016 02:21 PM, Bjorn Reese wrote: > On 10/27/2016 08:26 AM, Antony Polukhin wrote: > >> - Whether you believe the library should be accepted into Boost >> * Conditions for acceptance > > I withhold judgement for now to give the author an opportunity to > respond. At the moment I am unsure whether to vote for conditional > acceptance or rejection. I vote for conditional acceptance, provided the following conditions are met: 1. The interface style must be redesigned such that: a. There is a clear separation between arguments and attributes. b. Arguments and attributes should not be arbitrarily interspersed in the parameter list. c. A set of attributes can be build in advance, and reused for different child processes. d. Attributes can be optionally to the attribute-set. 2. There should be no I/O deadlocks by default. 3. An Overview section that introduces the concepts and the various parts of the library must be added to the documentation.

1 0

[process] Formal Review starts today, 27 October
by Antony Polukhin 05 Nov '16

05 Nov '16

Dear Boost community, The formal review of Klemens David Morgenstern's Process library begins today, 27th October and ends on 5th November. Process is a C++11 library to manage system processes. It can be used to: * create child processes * setup streams for child processes * communicate with child processes through streams (synchronously or asynchronously) * wait for processes to exit (synchronously or asynchronously) * terminate processes Full documentation with examples and tutorial is available at http://klemens-morgenstern.github.io/process/index.html Stable source codes for review are available at https://github.com/klemens-morgenstern/boost-process/tree/boost_review Latest source codes available at https://github.com/klemens-morgenstern/boost-process We encourage your participation in this review. At a minimum, kindly state: - Whether you believe the library should be accepted into Boost * Conditions for acceptance - Your knowledge of the problem domain You are strongly encouraged to also provide additional information: - What is your evaluation of the library's: * Design * Implementation * Documentation * Tests * Usefulness - Did you attempt to use the library? If so: * Which compiler(s) * What was the experience? Any problems? - How much effort did you put into your evaluation of the review? We await your feedback! -- Best regards, Antony Polukhin

10 38

[process] Formal Review
by Raphaël Londeix 04 Nov '16

04 Nov '16

> > We encourage your participation in this review. At a minimum, kindly state: > - Whether you believe the library should be accepted into Boost > YES > * Conditions for acceptance > The interface used to spawn processes should be reworked. > - Your knowledge of the problem domain > I have implemented some process wrappers many times. I have used the last release of Boost.Process (0.5 I think?). > - What is your evaluation of the library's: > * Design > I think the similarity with the boost::thread api is a great choice. The lifetime of process is handled by small functors that will be called at specified times. I think is a very good design as it allows one to write small "plugins" in order to affect the setup or the tear down of a process. The integration with Boost.Asio follows the previous choice of using boost::thread as a reference, i.e. the child process is not the resource you read from/write to, you need some separate stream objects whose lifetime are tied to the child process: struct MyProcess { bp::async_pipe std_in; bp::async_pipe std_out; bp::async_pipe std_err; bp::child child; }; I haven't seen a way however to be asynchronously notified when a child exited, so I'm not sure how one would handle gracefully multiple async_read/write while the process could exit at any time. The choice of having a nice interface to spawn processes have been discussed already, but I still think that the library should aim to have the simplest and lowest level interface. IMO, C++ users do not need to have a nice and expressive syntax to spawn process, I cannot imagine a code base where many processes are launched at many places in the code. In other words, I think that either you rarely need to spawn a process (a hack for example), or your job is to spawn processes (a shell, a make clone, ...). In both cases, you will spawn processes in very few places in the code. Moreover, this interface style has an implication on build times, errors seen at build time and makes it hard to use in cases where your program needs to spawn processes based on runtime decisions. if one decided to implement a shell that supports redirections, pipes, environment modification and background jobs, they would certainly need to make runtime adjustments to the arguments given to child constructor. As of today, this has to be worked around by the user ; I think this should be the primary interface. A suggestion has been made to have a more idiomatic C++ construction, I would extend it with a specific options type: auto options = bp::options().args("gcc"); if (verbose) options.append_arg("-v"); if (ignore_errors) options.std_err.redirect(bp::null_device); bp::child child(io_service, std::move(options)); This way you make it impossible to modify the options after the child process creation. But that's just an idea :) * Implementation > The implementation is clear, but the support for the aforementioned constructor interface "leaks" into the implementation everywhere. As the previous version of Boost.Process, the different OS support is nicely separated and each stage of the process creation is clearly implemented in its own header. > * Documentation > The asynchronous section feels a bit short given the many possibilities the interface gives us. > * Tests > Didn't check > * Usefulness > I think it would be very useful to have something we can rely on instead of our own half baked implementations. > - Did you attempt to use the library? If so: > * Which compiler(s) > g++4.8 using c++11 > * What was the experience? Any problems? > I had many problems and decided to fallback on popen(). The first kind of problems were some compile errors. But the author kindly stated that I needed the HEAD of boost (which is not an option for me) but I found a workaround by not using some niceties of the interface: https://github.com/klemens-morgenstern/boost-process/issues/20 https://github.com/klemens-morgenstern/boost-process/issues/21 As said previously, for someone not used to boost.fusion, the errors are hard to read. The other problem that made me give up (temporarily) on using Boost.Process was the fact that asynchronous reads where unreliable when more than one process is spawned. https://github.com/klemens-morgenstern/boost-process/issues/22 This issue has been closed despite the fact that it is not fixed. - How much effort did you put into your evaluation of the review? > I'd say half a day counting this mail and the testing I made. Cheers, -- Raphaël Londeix http://hotgloupi.fr

2 1

[process] Formal Review
by Klaim - Joël Lamotte 03 Nov '16

03 Nov '16

Boost.Process Review ==================== ...At a minimum, kindly state: - Whether you believe the library should be accepted into Boost * Conditions for acceptance YES it should be accepted, though documentation need improvements. - Your knowledge of the problem domain I implemented naive and less naives ways to manage different processes in the past. I currently work on libraries which are all about multi-process communication and we have a poor-man's process management (launching, tracking, etc.) api that I have to maintain on Linux, Windows and MacOSX. I also have home projects which are organized as client-server with master-slave relationship, clients spawning servers and pinging them to keep processes alive mutually (using network communication). So basically, I implemented basic process management libraries, but I'm more a process management library user, in all my projects, paid or not. I have been exposed to a lot of problems related and solved a lot of them. I also used the last 2 Boost.Process versions proposed to Boost (though I voted no for these ones) but in cases where I just needed a cross-platform way to launch an arbitrary executable (no tracking nor communication). You are strongly encouraged to also provide additional information: - What is your evaluation of the library's: * Design The weird choice of variadic arguments in launching functions have been explained so I'm more at ease with it, though maybe ways to simplify it could be attemted in the future. I like that the design of boost::process::child follow, where it make sense, the design of std::thread. Both could be considered as execution resource, so it's fine to me that they share similar interface phylosophies of design (like spawning on construction and move-only). I would love it if there was a way (even slow) to track processes lifetime for processes that I don't have code for. Though it's ok for me if it's a future extension. I like the minimalism of the scope, though useful extensions could be added later. * Implementation I briefly eyescanned the code, looked more closely at the termination trigerriung code because of the discussion we had on termination reqest vs kill. I don't bother much about performance for this library as launching processes is rarely done in a tight high-performance loop. * Documentation I did a report of the documentation a few weeks or months ago, I do not know if changes have been made since. Anyway I think the documentation should have a part focused on explaining the concepts because even if most programmers knows what a process is, all related concepts (like what is the "environnement") are so specific to processes that they are often ignored as long as naive usage works. So as soon as the user need more specific usage, most, I believe, will not know what word to use to find the right concept. I think this library documentation should follow something similar to Boost.MSM which introduce to the basic concepts, though without walls of text. * Tests I didn't have time to try it yet in real applications, but so far didn't find any problem in basic usage. * Usefulness It is very useful for almost all my projects. - Did you attempt to use the library? YES If so: * Which compiler(s) Visual Studio 2015 for now. * What was the experience? Any problems? I only did minimal testing, nothing deep, no problems. - How much effort did you put into your evaluation of the review? Several hours diffused through several months. Joël Lamotte

1 0