November 2013 - Boost - lists.preview.boost.org

Thanks to everyone for all the work on 1.55.0
by Marshall Clow 11 Nov '13

11 Nov '13

I just wanted to take off my "release manager" hat and say, personally, thank you to all the people who tried out the RCs/Betas/etc, reported bugs, and suggested fixes. You were all a big help in making the release. -- Marshall Marshall Clow Idio Software <mailto:mclow.lists@gmail.com> A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait). -- Yu Suzuki

1 0

[release][1.55.0] Visual C++ known bugs
by Daniel James 11 Nov '13

11 Nov '13

Hello everyone, I'm compiling a list of known issues in boost with Visual Studio 2013/Visual C++ 12. I've made a quick start at: http://www.boost.org/users/history/version_1_55_0.html#version_1_55_0.known… Please let me know if there's anything I should add. Also, if available, tested patches against the release branch or release candidate would be very useful (I don't have the compiler myself, so I can't do that). If you can, please make them using subversion from the root of the repository. thanks, Daniel

5 9

New Version (0.4) of Boost.Application (Aspect Based) available to comments (RFC).
by Renato Forti 11 Nov '13

11 Nov '13

>> How is that supposed to work in Windows? (Does it need to?) Well, a simple analysis, it seems to be specific in POSIX, we do not have anything like it in WINDOWS. (I do not speak for sure) >> Its sensible to expose 'something' that can be integrated to native event loops, but the ideal choice under Windows is probably an event, not a pipe descriptor. I agree. I'm thinking of adding this aspect only in posix, Then if user want use it need add it and use it only on UNIX, e.g.: class myapp { public: int operator()(application::context& context) { #if defined(BOOST_ BOOST_POSIX_API) boost::shared_ptr<application:: selfpipe > selfpipe = context.get_aspect<application:: selfpipe >(); select(...,selfpipe->readfd(),...); #endif return 0; } }; // main int main(int argc, char *argv[]) { myapp app; application::context app_context; #if defined(BOOST_ BOOST_POSIX_API) app_context.add_aspect<application:: selfpipe>( boost::make_shared<application:: selfpipe>()); #endif return application::launch<application::common>(app, app_context); } Comments, Ideas, please! >> I think some analysis is needed. I need help in this subject, if any one with great experience on both sides can propose something (shine a light), it would be of great value! Thanks for help

2 2

Data Structure and Algorithms yet to be inplemented
by Manmohit Rekhi 11 Nov '13

11 Nov '13

Hi, I would really be grateful if could get to know 1)which Data Structures and Algorithms are yet to be implemented? 2)How can i get started on them? 3)Are there any constraints on the libraries i can use to implement them? Thanking you Manmohit Rekhi

1 0

[UUID] Document typo
by Naked Short-Selling 11 Nov '13

11 Nov '13

http://www.boost.org/doc/libs/1_55_0/libs/uuid/uuid.html#Nil I believe is_null() really meant "is_nil()" here...

2 1

[git][ Last merge before git conversion.
by Vicente J. Botet Escriba 10 Nov '13

10 Nov '13

Hi, I would like to know if there will be some time (~1 week) to merge the new features and bug fixed in the release branch before svn is closed for git migration. Note that the provisional calendar was: 8 November : release 1.55 15 November: Provisional: Close subversion for git conversion. Best, Vicente

2 1

[intrusive] possible workaround for vc12 compile errors
by Daniela Engert 10 Nov '13

10 Nov '13

Hi, I've attached a patch to ticket #9332 "has_member_function_callable_with.hpp compile error on msvc-12.0" (https://svn.boost.org/trac/boost/ticket/9332) This patch seems to fix the problem. My far-from-perfect understanding of C++ seems to indicate, that the root cause is in line 201: template< class F , std::size_t N = sizeof((boost::move_detail::declval<F>(). BOOST_INTRUSIVE_HAS_MEMBER_FUNCTION_CALLABLE_WITH_FUNCNAME (), 0))> struct BOOST_PP_CAT(zeroarg_checker_, BOOST_INTRUSIVE_HAS_MEMBER_FUNCTION_CALLABLE_WITH_FUNCNAME) the sizeof expression is evaluated with an unknown type U in declval<U>(). Ciao Dani

2 1

Re: [boost] sqlpp11: SQL for C++
by Roland Bock 10 Nov '13

10 Nov '13

On 2013-11-10 15:58, niXman wrote: > Roland Bock 2013-11-10 12:15: > >> No, sqlpp11 itself only depends on a decent C++11 compiler and matching >> STL. > > I think I can write a code generator using boost.preprocessor. But I > need the documentation of DDL, of course. > I was talking about SQL's DDL (http://en.wikipedia.org/wiki/Data_definition_language) Unfortunately every vendor seems to be using some additions, so that the task is not too simple in all generality, I guess... But using the preprocessor to generate the table types could be a very nice tool in some cases. For instance if you are using in-memory databases with Sqlite3. As a trivial byproduct it could also produce the create table string, so that I would not have to maintain that in two places. Also for such a purpose, it would make sense to restrict the DDL to a common subset. With larger tables, I guess I'd stick with my current approach though: running a separate code-generator as part of the build process. The main benefit is that I get clearer error message pointing to C++ code that exists outside the compiler memory in case the code generator produces incorrect output or I use the generated code in the wrong way, etc. Another benefit of using external code generators is string handling. In our company we use CamelCase for classes, camelCase for members and underscore separated compounds for SQL names. An external code generator can easily translate, of course, but the preprocessor is not equipped for string manipulation, afaik. But again, I can imagine a nice usage of a prepocessor-based code generator for in-memory databases with no external representation. Regards, Roland

1 0

[utility] Any interest for a macro which automatically creates default move-semantics, comparisson etc for a class
by Viktor Sehr 10 Nov '13

10 Nov '13

I've created a macro which automatically creates: - move constructor\assignment - swap member function - operator== and operator!= - operator< - assignment operator without side-effects if it throws. Example usage: class Foo { public: Foo() {} DEFAULT_EVERYTHING(Foo, ( (int)(a) ) ( (std::vector<int>)(b) ) ( (float)(c) ) ); }; Any interest, suggestions or ideas? Ofcourse it might be split into several macros depending on what operators the class requests (if it's a move-only class etc), and perhaps an extra parameter enable if a member is up for sorting and comparison. Code available here: https://www.dropbox.com/s/dhkxj94rrh83az0/DefaultEverything.h

3 10

[containers] Are there flat_map/set and stable_vector proposals at work?
by Klaim - Joël Lamotte 09 Nov '13

09 Nov '13

All is in the title: I would like to know if flat_map, flat_set and stable_vector have been proposed for inclusion in the C++ standard library, or if there are works going on to write such a proposal. I find that I use a bit more flat associative containers than tree-based standard associative containers in all my recent projects, and I think that having option in the standard to choose one or the other would be very useful for a lot of devs.

5 20