September 2014 - Boost - lists.preview.boost.org

Re: [boost] Finite state machine
by arcolight 07 Sep '14

07 Sep '14

> Hi, > I don't know if you're using msm or statechart but msm uses no dynamic > memory, except if you choose to use the event queue or deferred events. And > even this can be avoided by providing as template argument you own queue. > If you don't require these features, you can disable them. > > Cheers, > Christophe I'm using MSM. Yes I tried this, but I could not do so. Perhaps, I doing something wrong. Problem in that i've no OS, and even no full C library, I'm using newlib-nano. I was trying to disable everything, that I can, but this didn't help. I understand, that this situation is special and boostmust count on something. May be you can advise me how to use (and, may be,build?)boost with arm-none-eabi compiler.I've read docs and trying to build boost for arm-none-eabi, but compilation fails. Do I understand correctly, that MSM doesn't need to compile something - this is only .hpp files and it doesn't use libs, so I can try to use it without compilation? -- arcolight

2 1

[git][align][litre] git submodule update failure.
by Mostafa 07 Sep '14

07 Sep '14

$ git submodule update libs/align fatal: reference is not a tree: d32353ea7d39c0237202e3e460f3248f226db274 Unable to checkout 'd32353ea7d39c0237202e3e460f3248f226db274' in submodule path 'libs/align' $ git submodule update tools/litre fatal: reference is not a tree: 564d4d8d30b7e03ac5e25d78e14d2c19fa321c83 Unable to checkout '564d4d8d30b7e03ac5e25d78e14d2c19fa321c83' in submodule path 'tools/litre' According to this: http://stackoverflow.com/questions/13425298/git-submodule-update-fatal-erro… , this is most likely a maintainer's bug. From the above: "This is the most common problem with submodules. The commit that you are on in the outer repository has a reference to a commit in the submodule that someone did not push up yet. It's a dependency problem. Always push from the inside-out. This is most likely not something you did wrong, but someone else that's working in the repository. They were sloppy and forgot to issue push on the submodule and just pushed the containing repository. Stuff works on their machine because they made the change and those commits exist there." Is this true?

3 3

Re: [boost] Finite state machine
by arcolight 07 Sep '14

07 Sep '14

> Are you using Statechart or MSM? > -- Michael Caisse ciere consulting ciere.com I tried to use both. -- arcolight

1 0

Re: [boost] Finite state machine (arcolight)
by arcolight 07 Sep '14

07 Sep '14

> This is something that would be very interesting to me, but I wonder if the > memory allocation can be factored out in the current boost library by using > allocators or something. I'm not terrible familiar with the libraries > internals, but I'd be willing to take a look at my next free cycle. To be > clear, is there another state machine lib in boost besides > boost.statechart? That's the only one I'm familiar with from a user > perspective. > > -Greg Yes, custom allocator doesn't solve the problem, especially because current FSM-s includes many other libraries. In any case, I can't afford to use even custom allocator in my project, if I would use it, I must have sufficient size heap, it's risky in embedded system. I already have third-party code, that uses heap, and I couldn't do anything with it. :) Another FSM implementation - Boost.MSM(Meta state machine), I tried to use it in different ways, but i've failed. MSM based on MPL boost library. If you interested in code, I have small repo on github: https://github.com/arcolight/ecl, there are several small classes, one of them, in fsm.hpp - state machine. There are simple test with cd-player like in boost lib.

1 0

Finite state machine
by arcolight 07 Sep '14

07 Sep '14

Hi all. I have a question about participation in boost libraries. In my embedded project I'm using finite state machine and I tried to use one of the boost implementations of FSM, but they are too big and complex. Main disadvantage of them is usage of dynamic memory allocation, that I can't afford on my CPU and many dependencies, that I can't use on my CPU. CPU is ARM Cortex-M4 with 128K memory, I've no operating system in my project. I have write my own FSM implementation without dynamic memory and dependencies, it's more simple, than boost libraries, but, of course, not so powerful. I've used C++11 standard, specifically variadic templates. My question - Is there some interest in such FSM implementation for boost developers? I'll be very grateful for advice. P.S.: I'm sorry for my english, it's not my native language.

4 3

Autolinking broken in master
by Agustín K-ballo Bergé 06 Sep '14

06 Sep '14

There's a report that autolinking is broken for a fresh recursive checkout of master, due to b2 generating 1.57 libraries while boost/version.hpp still reports the version as 1.56. This commit setting BOOST_VERSION in Jamroot made it to master: https://github.com/boostorg/boost/commit/d9cf98c2074cf8a27017170375c5c825e4… While the corresponding commit for boost/version.hpp didn't: https://github.com/boostorg/config/commit/34e1c7416be2ede7d86f3dcc093a84580… Regards, -- Agustín K-ballo Bergé.- http://talesofcpp.fusionfenix.com

1 0

[iterator] Merge to master
by Andrey Semashev 06 Sep '14

06 Sep '14

Hi, There were lots of changes to Boost.Iterator develop in the last few months. The tests look green for quite some time now. Could these changes be merged to master? I need those changes to merge next/prior in Boost.Utility.

2 1

sqlpp11, 3rd iteration
by Roland Bock 04 Sep '14

04 Sep '14

Dear Boosters, Quite a bit has happened since last I reported about sqlpp11 in this forum [1,2]. I have incorporated a lot of the feedback you gave me, hopefully bringing the library closer to a reviewable state. Source: https://github.com/rbock/sqlpp11 Doku: https://github.com/rbock/sqlpp11/wiki (not at all formal yet) I am hoping for more feedback both here and live at CppCon (http://sched.co/1r4lue3) Here are some noteworthy changes: _Restructuring:_ The code has undergone a major restructuring, leading to a much simpler and thereby much more flexible way to define the structure of SQL statements. As a result, it would now be quite simple to add vendor specific features sqlpp11 connector libraries. Dominique Devienne mentioned array binding for inserts, Johan Baltié mentioned hierarchical searches (both features of Oracle). If you are interested in writing a connector and adding such features, please let me know, I'll walk you through it. _NULL handling:_ Enabled by the restructured code and spurred by the library quince by Michael Shepanski, sqlpp11 can now calculate which result fields can or cannot be NULL. Speaking of which, handling NULL for result values has been discussed a lot. The library now has compile-time configurable behavior, you can choose between an std::optional-like interface and mapping NULL to the trivial value of the result type, e.g. 0 for numbers or "" for strings, see also https://github.com/rbock/sqlpp11/wiki/NULL _Connectors:_ Matthijs wrote a new postgresql connector from scratch, see https://github.com/matthijs/sqlpp11-connector-postgresql The sqlite3 connector received a bunch of corrections, thanks to dirkvdb _Compilers:_ Known to compile with clang>=3.1 and gcc>=4.8. I currently cannot test with MSVC or Intel or others, but if you happen to have the latest top notch version of any of those, I'd be interested in what they have to say about sqlpp11. Cheers, Roland [1]: http://lists.boost.org/Archives/boost/2013/11/208388.php [2]: http://lists.boost.org/Archives/boost/2014/02/211206.php

7 43

[move][unique_ptr] c++14 unique_ptr comes to town
by Ion Gaztañaga 03 Sep '14

03 Sep '14

Hi to all, We've discussed about a boostified version of std::unique_ptr several times in later years. I decided we could not continue with this hole in the Boost toolset and implemented in Boost.Move a unique_ptr that follows the C++14 interface. Commit here: SHA-1: e1da7c5ca1162841f643331af891a9a730eba118 http://github.com/boostorg/move/commit/e1da7c5ca1162841f643331af891a9a730eb… This implementation is also usable from C++03, and apart from the general limitations of the Boost.Move machinery, it should be really uesful for C++03 users. I would like to make this implementation the official boost::unique_ptr implementation, but until the community approves this, it's in boost::movelib namespace (boost::move it's a function so I needed to tweak the namespace name). I would like to thank Howard Hinnant for his great public implementation of the emulated unique_ptr published here: http://home.roadrunner.com/~hinnant/unique_ptr03.html I've ported and refactored Howard's extensive testsuite to implement boost test cases. Really useful to detect many implementation errors and details. I've found unique_ptr much harder to write than I expected. The standard text is really detailed and helps implementors, but things like allowed conversion between pointers and deleters and several ICEs and different SFINAE parties on many compilers made the work quite entertaining. The API is taken from the draft n3797 and the implementation has some differences with the standard. The main reason for this is that I wanted to make unique_ptr as lightweight as possible avoiding external dependencies, as it's interesting to implement idioms like PIMPL. Main attributes and differences with the standard: -> Implemented in two headers: <boost/move/unique_ptr.hpp> and <boost/move/make unique.hpp>. -> <boost/move/unique_ptr.hpp>: Boost.Move is the only heavyweight dependency. Not even that, as unique_ptr.hpp only uses boost/move/utility_core.hpp, that does not depend on any external library excepct BoostConfig and StaticAssert. No TypeTraits, MPL and others. Only Boost.Config, BOOST_STATIC_ASSERT and BOOST_ASSERT are used in the implementation. No standard library headers used (no pointer_traits, memory or other utilities). -> <boost/move/make unique.hpp>: Boost.PP is used in compilers without variadic templates. This is the main reason to have a separate make_unique.hpp header and avoid pulling Boost.PP for c++03 users that only want to have C++11 unique_ptr. -> unique_ptr's "operator <" is implemented as x.get() < y.get() instead of std::less<CT>(). std::less requires <funcional>. std::common_type requires C++11 standard library or some partial emulation based on Boost.TypeOf. I don't think 99,99% of users will notice the difference. For raw pointers, in all our platforms comparisons provide strict weak ordering. If a smart pointer writer specializes std::less for its smart pointer, it should be a law against that. -> It's assignable from literal 0. I found some problems in some compilers trying to write a constructor for that. -> nullptr support is conditional on !defined(BOOST_NO_CXX11_NULLPTR) but I think really should have a Boost nullptr for C++03 compilers. Let's put Meyer's nullptr (http://en.wikibooks.org/wiki/More_C%2B%2B_Idioms/nullptr) in Boost for BOOST_NO_CXX11_NULLPTR compilers. -> unique_ptr<T[]> is convertible to other types when conversion is considered safe. This means that unique_ptr<T[]> is convertible to unique_ptr<const volatile T[]>. Conversion from T[] to U[] is considered safe if remove_cv<T> is the same type as remove_cv<U> and unique_ptr<T[]>::pointer is convertible to unique_ptr<U[]>::pointer. This is a feature that might go to C++17 (http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-active.html#2118) Safe conversions in reset() and constructors from pointers are also implemented in unique_ptr<T[]>. I tested the implementation in GCC{4.4, 4.6, 4.7, 4.8} Clang3.4 and MSVC{7.1, 8.0, 9.0, 10.0, 11.0, 12.0} in 32/64 bits, debug/release and C++03/C++11 modes (where applicable). The class documentation is generated with Boost.Move documentation and the most part of it is taken from the standard. I haven't added a tutorial to Boost.Move but I plan to it based on Howard's c++03 unique_ptr tutorial. I hope you find this addition to Boost.Move useful. If you think the implementation is good enough to be the official boost::unique_ptr implementation living in Boost.SmartPtr, porting should be easy, except for the documentation part. That would require either combining somehow handwritten and Quickbook produced documentation or writing new docs for unique_ptr. Best, Ion

6 30

Default documentation settings
by Glen Fernandes 03 Sep '14

03 Sep '14

What is the value of chunk.section.depth used, for those Boost libraries whose documentation is supplied only in QBK form (not the generated HTML from the QBK)? It appears to be 1, i.e. my sections-within-sections are not broken into separate pages. If so, could we make the default <xsl:param>chunk.section.depth=2? Glen

2 4