- Boost - lists.preview.boost.org

[atomic] Support for specialized instructions
by Andrey Semashev 14 May '14

14 May '14

Hi, I've been contemplating the idea of adding support for some specialized atomic instructions to Boost.Atomic. Since I'm only familiar with x86 architecture, my current candidates are: * Increment/decrement with a check for zero result (lock inc/dec). The decrement is the most essential of the two because it would be useful for reference counters of various kinds. The increment is mostly beneficial in terms of code size, and it probably doesn't need the check for zero in most cases. * In-place logical operations (lock and/or/xor). These are useful in cases when the code needs to operate on flags or bit sets but doesn't need the result of the operation. Optional variants can be provided with a test for zero result. * Bit set/reset operations (lock bts/btr). The use cases are similar to the logical operations but the advantage is that the previous value of the altered bit can be returned. BTW, these operations are needed in Boost.Sync to implement mutexes on Windows. All these operations can be implemented through the standard atomic<> interface, and that's what would be done on platforms without support for the specialized instructions. For now the idea is to add special member functions to atomic<>, when used with arithmetic types. I was thinking about providing free functions as well, although I think that the complete set of the standard atomic functions would be required for that. Comments? Opinions?

4 6

[atomic] Major rewrite landed in develop
by Andrey Semashev 14 May '14

14 May '14

Hi, I've merged a major rewrite of Boost.Atomic into develop. The changes are quite massive to describe in a few words, but the most significant ones are: - The interface of atomic<> template and atomic_flag are now completely decoupled from the atomic operations implementation (the backends). It is now more consistent across all platforms and not duplicated everywhere. - atomic_flag and fences are now available through separate headers atomic_flag.hpp and fences.hpp. These headers are still included by atomic.hpp for backward compatibility. - The definition of configuration macros BOOST_ATOMIC_XXX_LOCK_FREE is now available through a separate header capabilities.hpp. Users are able to include it and test these macros before including the rest of Boost.Atomic. Most macros are now defined more correctly based on the sizes of the types (i.e. BOOST_ATOMIC_LONG_LOCK_FREE now actually depends on the size of long and is equivalent to BOOST_ATOMIC_INT32_LOCK_FREE or BOOST_ATOMIC_INT64_LOCK_FREE). - Added support for 128-bit atomic ops with MSVC. The support is always enabled for x86_64. There are some ancient AMD chips that don't support cmpxchg16b, users can define BOOST_ATOMIC_NO_CMPXCHG16B if they want to target those. - Added support for BOOST_ATOMIC_FLAG_INIT, although it is optional and atomic_flag is still zero initialized by default. I did not find a way to implement it and keep compatibility with C++03. - Improved ARM backend with gcc (tested for compilation). Now it supports 64- bit atomics. All operations are written in ll/sc manner instead of a CAS loop. - Added a separate ARM backend for MSVC (tested for compilation). Some support for ARM was added to develop earlier but the new implementation is better optimized (it uses appropriate intinsics wrt memory ordering). - Added support for 64-bit atomic ops for SPARCv9 with gcc (completely untested). - Added support for DEC Alpha backend with gcc (completely untested). It is based on the dead and incomplete code that was present in Boost.Atomic before. - Removed support for IA64 on Windows. Not sure if it is needed, but it can be added relatively easily. There are many more changes. Lots of code duplication removed, bugs with arithmetic overflow fixed and things I don't remember. This new version is the foundation for further improvements that are needed for Boost.Sync. In particular, I plan to add new operations (btr/bts, inplace logical operations) and maybe free atomic functions. You can see in the above list that much of the code is not properly tested, and I have little-to-none experience with some of the platforms. To my knowledge, there are no non-x86 regular testers. Therefore I'd appreciate if people reviewed and tried the code. Please, report any feedback, good or bad. Thank you.

1 0

[graph] Fix for directed_graph::swap
by Ben Pope 14 May '14

14 May '14

https://github.com/boostorg/graph/pull/5 Ben

1 1

compressed_pair breaks with multiple instances of empty types
by Hite, Christopher 14 May '14

14 May '14

I've been abusing compressed_pair to build a compressed_tuple. I'm pretty sure this bug can be fixed using intermediate types on the hierarchy. struct Empty1{}; struct Empty2{}; int i=sizeof( boost::compressed_pair<Empty1, boost::compressed_pair<Empty1, Empty2> >); //boom int j=sizeof( boost::compressed_pair<Empty1, Empty2>); //ok int k=sizeof( boost::compressed_pair<Empty1, Empty1>); //ok int h=sizeof( boost::compressed_pair<std::string, boost::compressed_pair<std::string, Empty2> >); //ok cc1plus: warnings being treated as errors In file included from ..../boost/boost_1_50_0/include/boost/compressed_pair.hpp:21, from /home/w461845/src/edgamm/src/lib/libLogger/include/libLogger/LogChain.hpp:4, from OLogTest.cpp:1: ..../boost/boost_1_50_0/include/boost/detail/compressed_pair.hpp: In instantiation of "boost::details::compressed_pair_imp<Empty1, boost::compressed_pair<Empty1, Empty2>, 3>": ..../boost/boost_1_50_0/include/boost/detail/compressed_pair.hpp:348: instantiated from "boost::compressed_pair<Empty1, boost::compressed_pair<Empty1, Empty2> >" OLogTest.cpp:112: instantiated from here ..../boost/boost_1_50_0/include/boost/detail/compressed_pair.hpp:73: error: direct base "Empty1" inaccessible in "boost::details::compressed_pair_imp<Empty1, boost::compressed_pair<Empty1, Empty2>, 3>" due to ambiguity g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4) Chris Hite This email is confidential and subject to important disclaimers and conditions including on offers for the purchase or sale of securities, accuracy and completeness of information, viruses, confidentiality, legal privilege, and legal entity disclaimers, available at http://www.jpmorgan.com/pages/disclosures/email.

1 0

Re: [boost] Is there any interest in a library for actor programming? [preliminary submission]
by Charousset, Dominik 14 May '14

14 May '14

Lets hope the mail arrives this time ... boost(a)lists.boost.org [129.79.39.203] #<[129.79.39.203] #5.0.0 smtp; 5.4.7 - Delivery expired (message too old) [Default] 451-'4.3.5 Server configuration problem' (delivery attempts: 76)> #SMTP# ________________________________________ From: Dominik Charousset [dominik.charousset(a)haw-hamburg.de] Sent: Friday, May 09, 2014 14:18 To: boost(a)lists.boost.org Subject: Re: [boost] Is there any interest in a library for actor programming? [preliminary submission] On 09.05.2014, at 10:52, Jan Herrmann <jherrmann79(a)gmx.de> wrote: On 08.05.2014 19:30, Dominik Charousset wrote: >> Hi all, >> >> >> attached you'll find the first preliminary submission for Boost.Actor as ZIP file along with the Manual in PDF version. You currently still need CMake, a Boost.Jam build file will follow. The project is hosted at GitHub:https://github.com/Neverlord/boost.actor > > As far as I understand a library with LGPL can not be part of boost (see www.boost.org/development/requirements.html ). So the it should be changed. Yes, we would switch the license to the Boost License should it become accepted. >> To state the obvious: we [1] took our time, but not because we are not taking the boost submission serious. In fact, the opposite is true. We did refine libcppa (which is the base for Boost.Actor) over the last months based on community feedback - feedback from this mailing list, last year's C++Now, GitHub, and libcppa's mailing list. However, we are not just trying to lift our library into Boost. We want Boost.Actor to fit into Boost as good as possible - and we did significant changes to the API (see below), which also took time. Furthermore, we think performance is key when providing a native actor system. Boost.Actor/libcppa has now reached a level of performance with the latest version that lives up to our expectations - and hopefully to the expectations of (future) users as well. A brief performance evaluation [2] is attached. The graphs still mention 'libcppa' instead of Boost.Actor. >> >> I hope the lack of Boost.Jam support is not too much of an issue - I thought it would be a good idea to post the preliminary submission before the C++Now (you can meet me there!), to get people time to play with Boost.Actor beforehand. >> >> >> *Differences between Boost.Actor and libcppa* >> >> If you don't have any experience with libcppa, you can simply skip this. The most visible change is the removal of any_tuple and cow_tuple. The COW tuple implementation of libcppa is in our opinion not general enough and is too intermeshed with the type system. Instead, Boost.Actor provides the class `message` that is an opaque container from a user's point of view. Also, the pattern matching has been reduced to a minimal DSL for defining message handlers. The separate DSL for guard expressions is gone completely (a similar functionality still exists though). Since our pattern matching only worked with messages/any_tuples, we don't see too much use of it outside of Boost.Actor. We are still hoping for native pattern matching in C++ (N3449 proposed this a while ago) in which case our DSL would become unnecessary anyways. As a side-effect of removing all these parts, Boost.Actor has become a streamlined version of libcppa that is easier to grasp and that will be easier to mai > ntain i > n the long run as well. >> >> >> *Boost.Actor depends on...* >> >> variant, optional, none, string_algorithms, intrusive_ptr, thread (currently as header-only dependency), context/coroutine (for cooperatively scheduled blocking actors). Some examples also require program_options. >> >> >> *Boost.Actor does NOT use...* >> >> (1) MPL. Boost.Actor uses a ludicrous amount of metaprogramming based on boost::actor::detail::type_list. I've tried to port it to MPL and wasted almost two days doing so. MPL is a great C++ library. Unfortunately, C++98, not C++11. Even after adding variadic helpers to reduce the amount of template instantiations, it would still explode and fail to compile. I've stopped after writing half the size of the current type_list utility functions and still wasn't able to reduce the amount of template instantiations significantly enough. That being said, I am looking forward to the GSoC project for MPL11. >> >> (2) Serialization. The first and most obvious problem I have with this library is the lack of a portable binary archive. String serialization is sufficient for a lot of applications, but the performance hit for both networking overhead and the string serialization itself is not acceptable in an actor system. Furthermore, simply adding such a binary archive would not be enough. The library does not give you a platform-independent representation of the type names, which is crucial. Also, one cannot create an instance of a type given only its type name and then deserialize its values from a network stream (I was thinking of ETI::key_register using the uniform names and ETI::find, but it only solves a part of the problem). Adding all of these capabilities on top of boost.serialization plus writing a portable binary archive (or maybe use something like EOS - http://epa.codeplex.com/) would probably result in more code and most likely in a more complicated design, i.e., would be > harder to maintain. Maybe I miss the forest for the trees here and some experienced boost.serialization user knows a straightforward way I don't see. However, maybe there is interest in extending boost.serialization. I'd be happy to work on patches for adding uniform names, etc. >> >> (3) Boost.Asio. There is definitely potential to port the low-level parts of the middleman to Asio, I think. These changes would be transparent to the user, though, since users really only use the broker and can forget about the rest. >> >> >> *Separating parts of Boost.Actor?* >> >> Vicente J. Botet Escriba suggested to separate some parts of library. Namely the constexpr atom() [the other parts he suggested were removed, btw]. Aside from that, there are some C++11 utility functions that I think would be useful for other developers as well. Namely int_list [3] which can be used to create indices for variadic types such as tuples to allow users to apply the values of a tuple to a functor with apply_args [4]. type_list [5] is a powerful tool too, but I think this functionality goes to MPL11 anyways. Something that might be worth lifting to the boost namespace is unit [6], which makes metaprogramming (for me) a lot easier by providing an easy way of lifting void to something useful and don't need to worry whether or not it might end up in a tuple or variant. MPL already has void_t, but it doesn't seem to be indented to be used the way I use unit. For binary serialization of floats, there is IEEE 754 packing/unpacking [6] that is based on http://beej.us/gu > ide/bgn > et/examples/pack2.c > > What is the difference from your int_list and std::integer_sequence (http://en.cppreference.com/w/cpp/utility/integer_sequence) ? The difference is: integer_sequence didn't exist when I wrote int_list. But it's good to know that there's a standard way to do it in C++14. Thanks for the hint. By the way, is there any policy regarding C++11 in Boost? Should we just skip it and directly aim for C++14? The code base is currently C++11 compliant. Switching to C++14 would make a thing or two easier (i.e. I could remove some code). >> Sorry for the wall of text, the Mail got longer and longer somehow ... >> >> >> >> [1] "We" are a working group at the University of Applied Sciences Hamburg: http://inet.cpt.haw-hamburg.de/ >> >> [2] The sources for all benchmark programs can be found here: https://github.com/Neverlord/cppa-benchmarks, please keep in mind that the evaluation uses libcppa, so the API used in the benchmarks is different. >> >> [3] https://github.com/Neverlord/boost.actor/blob/a0cab41e2946df7d751c59dfead0c… >> >> [4] https://github.com/Neverlord/boost.actor/blob/ea53a515e09a98baa173ad379127c… >> >> [5] https://github.com/Neverlord/boost.actor/blob/58d6177837788b2e54b8547531bda… >> >> [6] https://github.com/Neverlord/boost.actor/blob/644a47760d6b55eef438a81d7bcdc… >> >> [7] https://github.com/Neverlord/boost.actor/blob/ea53a515e09a98baa173ad379127c…

1 0

Re: [boost] Is there any interest in a library for actor programming? [preliminary submission]
by Dominik Charousset 14 May '14

14 May '14

> On 11 May 2014, at 09:36 pm, Larry Evans wrote > > Hi Dominik, > > On 05/08/14 12:30, Dominik Charousset wrote post > http://article.gmane.org/gmane.comp.lib.boost.devel/250747 > including: > [snip] >> >> [3] https://github.com/Neverlord/boost.actor/blob/a0cab41e2946df7d751c59dfead0c… > > However, I have a question about the implementation. > > https://github.com/Neverlord/boost.actor/blob/a0cab41e2946df7d751c59dfead0c… > > uses a recursive call using a List<typename ...Ts> which is 1 shorter > than the argument list, List<T0, Ts...>. > Why not just pass sizeof...(T0, Ts...): > > http://en.cppreference.com/w/cpp/language/sizeof... > > and keep adding indices while decreasing this sizeof value, > as done here: > > http://svn.boost.org/svn/boost/sandbox/variadic_templates/boost/mpl/package… > > in the package_range_c template class where: > > Start value would be 0 > and Finish value would be sizeof...(T0, Ts...) > > It achieves the same result, but without producing unneeded > instantiations. For example: > > package_range_c<std::size_t,0,sizeof...(mpl::vector<T0,T2,T3>>::type > package_range_c<std::size_t,0,sizeof...(mpl::vector<S0,S2,S3>>::type > > would produce 1 instantiation of: > > package_range_c<std::size_t,0,3> > > whereas the int_list.hpp code would, IIUC, produce the same > number of instantiations for the T's and S's. > > Wouldn't using the sizeof... method be easier to understand > as well as produce few instantions? Reviewing the code again, I think there's really no good reason to recurse over Ts. Doing the unrolling in il_indices recursively with only two specializations is a simple design, but the instantiations of type_list are indeed unnecessary. Also, the dependency to type_list could be dropped when unrolling based on sizeof.... > -regards, > Larry > > NOTE: I tried to post to the newsgroup; however, apparently my > post got lost :( Let's hope the list is working again...

1 0

Boosr review of the 'convert' library starts tomorrow May 21.
by Edward Diener 13 May '14

13 May '14

The review of the Convert library starts tomorrow Monday, May 12 and goes through Wednesday May 21. The Convert library builds on the boost::lexical_cast original design and experience and takes those conversion/transformation-related ideas further. * to be applicable to a wider range of conversion-related use-cases, * to provide a more flexible, extendible and configurable type-conversion framework. The Convert library can be cloned from GitHub at https://github.com/yet-another-user/boost.convert. The library follows the modular-boost format. Just clone it to modular-boost/libs in the 'convert' subdirectory and run 'b2 headers' in order to create a link to its header file directory in the modular-boost/boost subdirectory. The library comes with documentation in its top-level index.html or doc/html/index.html file. The library is by Vladimir Batov and is a second reiteration with a greater focus of his library that was reviewed in the past. I am Edward Diener and I am again serving as the review manager of the library. Comments, questions, and reviews will all be welcome for the library. Please try to sum up your review by answering these questions: What is your evaluation of the design? What is your evaluation of the implementation? What is your evaluation of the documentation? What is your evaluation of the potential usefulness of the library? Did you try to use the library? With what compiler? Did you have any problems? How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? Are you knowledgeable about the problem domain? And finally: Do you think the library should be accepted as a Boost library? As always whether you do or do not feel that the library should be accepted into Boost please specify any changes you would like to see for the library to be better in your estimation.

4 3

[RFC] Modern CORBA C++ IDL Mapping and implementation
by Felipe Magno de Almeida 13 May '14

13 May '14

Hello Boosters, Where I work we are implementing an open source, modern, CORBA implementation and IDL mapping. We want it to support C++ and C++11 IDL mapping, but we also want to create another IDL mapping. This new IDL mapping is being experimentated in the following url https://github.com/expertisesolutions/mORBid/tree/concepts It still has some very problematic issues (like compilation-time of generated code) and of course it is still far from completion. But, we would like to hear suggestions about this approach. What we did was that an IDL interface is mapped to a typedef for boost::type_erasure (I found the idea quite cool for the reasons that I'll mention below) with the requirements defined in the IDL, e.g. its operations. The IDL also emits a model for this type_erasure which is the actual stub class and is called remote_reference. It also emits a proxy_reference model for the same type_erasure typedef which receives a std::reference_wrapper and proxies the calls, this is needed because type_erasure can't switch between by-reference and by-value in runtime. Why I believe this approach is better than current IDL to C++11 mapping: Because the IDL to C++11 mapping is too intrusive. You can't really use normal classes you have written for use as servants in CORBA because you have to inherit from Skeletons classes *and* you have to use the *exact* same signature as defined in the mapping. This is *way too restricted*. This is also inerent to the OO paradigm that CORBA uses. But, we can mimic the same features using type-erasure instead of directly using dynamic polymorphism. This gives a much greater freedom to users. It might be that Boost.type_erasure is not the best tool here because of compile-time and compiler memory usage, but the idea surely is. We still haven't seriously profiled boost.type_erasure yet, but we will do so as soon as possible. I think it might be very interesting to experiment with lite concepts and Boost.concept_check as well. Also, we created a EDSL for GIOP and IIOP mapping based on the boost.spirit infrastructure. We use that to treat alignment and endianness automatically and to use the same grammar for generation and parsing of IIOP messages. The message header grammar is defined like this: start = "GIOP" & octet('\1') /* GIOP major version */ & octet('\0') /* GIOP minor version */ & endianness [ octet(MessageType) /* message_type */ & raw_size(ulong_)[body] ] ; Which is compiled to IIOP parser or generator through boost.spirit infrastructure of make_terminal, make_directive and etc. This works splendid well and is really easy to use, extend, is statically checked and has great performance. Whoever feel interested, please take a look and comments are highly appreciated. Regards, -- Felipe Magno de Almeida

5 5

Re: [boost] Is there any interest in a library for actor programming? serialization + uniform names
by Andrzej Horoszczak 13 May '14

13 May '14

2 1

Lack of response to pull requests [resend]
by Beman Dawes 13 May '14

13 May '14

IMO it should be a Boost policy that anyone who submits a pull request deserves a response within a reasonable period of time. *What is a "response"?* GitHub pull requests have a "Comment" mechanism, so that's the best mechanism as it ensures reaching the submitter even they don't read the developer list. We also need to delineate some common responses, but that can wait until the overall policy is agreed upon. *What is a "reasonable period of time"?* This may be a bit of a bike shed discussion since whatever we start with may need adjustment based on experience. Would two weeks be a good starting point? *Who should step in if there is no response from the library maintainer?* The Community Maintenance Team IMO is the logical choice. A pull request that does not receive a timely response may be an early warning sign that the library involved does not have an active maintainer. The CMT folks are interested maintenance and already have processes to review pull requests. This would be an expansion of their mandate, and details need to be worked out such as what happens if the library maintainer is just busy at the moment. *What should the pull submitter do if there is no response within the reasonable period of time?* Send an email to the CMT pointing to the open pull request. (It would be nice if the submitter didn't have to do anything, but that implies the ability to monitor open pull requests. See "[GitHub] List open pull requests for all libraries?") Comments? --Beman

3 2