January 2013 - Boost - lists.boost.org

[log] v2 Initialisation, default sink behaviour and filtering documentation
by Mats Taraldsvik 08 Feb '13

08 Feb '13

Hi, I'm in the process of adopting a logging library, so the timing of the Boost.Log v2 "status report" - for me - could not have been better. :) This library is awesome, thank you for making it! -- I am in the process of moving beyond the default sink now, and had some issues. I tried to add the sinks after my logger object had been default initialized, and this caused some problems (partial streams/corruption, mostly). Currently I'm calling core->remove_all_sinks() to remove the default sink. Are there any "best practices" here? Perhaps it would be a good idea to add these "requirements" to the documentation, if they are not there already and I missed them? :) -- I also noticed that the severity filter examples are : logging::core::get()->set_filter ( logging::trivial::severity >= logging::trivial::info ); >From "Setting up sinks", and sink->set_filter(expr::attr< int >("Severity") >= 2); (i.e. comparing the Severity with an int) when moving to the more advanced section in "Sink frontends". Could you adjust one of the examples in "Sink frontends" to use the severity_level type (that the documentation introduces in the tutorial "Adding more information to log: Attributes") directly? For example: sink->set_filter(expr::attr< severity_level >("Severity") >= warning); Or explaining this in "Log record formatting". I had trouble understanding this at first, but I might be the only one :). Thought I'd mention it anyway. -- Regards, Mats Taraldsvik Norkart AS norkart.no

2 5

[Function] : Memory allocation upon bind()
by Harry van Haaren 06 Feb '13

06 Feb '13

Hi All, Summary: Boost::bind() allocates memory on the heap. In my situaion it is necessary to use manual memory pool instead of heap allocation, to conform with real-time timing requirements. (no malloc() calls allowed in real-time code) A solution might be writing a custom allocator for the boost::bind()'s data, and use a pre-allocated memory pool with the "placement" new( void* ) to write directly into it. I've found that there was a discussion on std::allocator support for std::bind and std::function here: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2308.html I'm not very familiar with the internals of bind() and Function in general, hence advice is appreciated. What is the status of boost::bind() and custom allocators? Is there a chance this could work? Am I missing some "easy" workaround or solution? Thanks for reading, -Harry

2 2

[function] Documentation suggestion (FAQ)
by Evan Driscoll 04 Feb '13

04 Feb '13

I just wrote up a longish email (excerpted below) that was basically asking for advice about how to work around the fact that I can't compare two function objects. However, I fortunately decided to take another look at the docs before sending, and realized that Boost actually supports what I *really* needed, which was to compare a function object to see if it points to a particular function so I could use a much faster implementation. In other words, I was doing void backing_func() {...} boost::function<void()> func(backing_func); void foo(function<void()> callback) { if (callback == func) { fast_foo(); } else { slow_foo(callback); } } but by changing the if statement condition to if (callback == &backing_func) everything is fine and dandy. My suggestion is that the FAQ entry at http://www.boost.org/doc/libs/1_52_0/doc/html/function/faq.html should be updated to point out that there IS a way to do this specific case. If everyone read the whole docs and remembered them perfectly this wouldn't be a problem, but of course those statements are probably true of exactly no one. :-) I stumbled across the FAQ entry, which I feel strongly suggests that what I want to do is impossible. Evan P.S. If you have suggestions for a better design, I'm open to suggestions. --------- I'd like input about how to handle a design. I'll describe what I'm doing; my question is what people would suggest. I have a class which stores a boost::function<> as a parameter so as to provide a hook to customize its behavior. I can give the actual code and description of what I'm doing if that will help, but I suspect it distract more than help, so I'll just make something up (untested): typedef boost::function<bool (int, int)> CallbackType; class MyClass { CallbackType callback; void foo() { ... if (callback(x, y)) { .... } } MyClass(CallbackType cb) : callback(cb) {} }; There are a couple generically useful behaviors, so the library predefines callbacks for those: bool behavior_a(int, int) {...} bool behavior_b(int, int) {...} bool behavior_c(int, int) {...} CallbackType behavior_a_callback(behavior_a), behavior_b_callback(behavior_b), behavior_c_callback(behavior_c); I have need to make 'foo' fairly general and support fairly general callbacks, but for one behavior (say behavior_a), I can use a substantially more efficient algorithm (log or constant time, instead of linear) inside foo(). What I WANTED to do was make the behavior_a(int,int) function just a dummy, then do void foo() { if (callback == behavior_a_callback) { fast_foo(); } else { slow_foo(); } } However, for reasons described at http://www.boost.org/doc/libs/1_52_0/doc/html/function/faq.html#id1617166, you can't compare functions with equality. Any thoughts as to how you'd do this? Evan

2 1

new proposal: order statistics tree
by Szymon Wojciechowski 03 Feb '13

03 Feb '13

Hi, I would like to propose to add new data structure - order statistics tree. Briefly, it is container which allows logarithmic inserting, searching and erasing as in set, but additionally it permits to access elements via numerical value - key order. I have some ready implementation with all appropriate functions (without their all overloaded forms) with gtest regression list, but the code isn't boostified. The regression passes on gcc 4.6.3 and vs2010. Is there anybody interested in? Best regards, Szymon Wojciechowski

7 21

[fixed_point] Request for interest in a binary fixed point library
by Vicente J. Botet Escriba 02 Feb '13

02 Feb '13

Hi, the recent discussion on the MultiplePrecission Arithmetic library has show that some people has its ow fixed point library. Is there an interest in a Boost library having as base http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2012/n3352.html? I have started a prototype http://svn.boost.org/svn/boost/sandbox/fixed_point (there is no doc yet but the basic goal is quite close to the N3352 C++ proposal). This prototype should allow you to play a little bit with. If yes, what would you like to be changed, added or improved in n3352? Next follows some design decisions that IMO need to be decided before hand. * Should integers and reals be represented by separated classes? * Should signed and unsigned be represented by separated classes? * Should the library use a specific representation for signed numbers (separated sign, 2-complement? Let the user choose? * Should the library provide arbitrary range and resolution and allocators? * Should the library be open to overflow and rounding or just implement some of the possible policies? and in this case which ones? * Should fixed_point be convertible to/from integer/float/double? * Could the result of an arithmetic operation have more range and resolution than his arguments? * Is there a need for a specific I/O? * is there a need for radix other than 2 (binary)? * Should the library implement the basic functions, or should it imperatively implement the C++11 math functions? Could a first version just forward to the c++11 math functions? * Should the library support just one of the know ways to name a fixed-point, a U(a,b), nQm, ...? Provide some ways to move from one to another? * Could expect the same/better performances respect to hand written code? * What should be the size used by a fixed_point instance? _fast? _least? Should the user be able to decide which approach is better for his needs? * Which should be the namespace? boost? boost/fixed_point? boost/binary_fixed_point? boost/bfp? ... * others you can think of. Please, replay if you are interested, have some experience in the domain, good ideas, .. Best regards, Vicente

6 44

[1.53.0] Release candidates available
by Marshall Clow 02 Feb '13

02 Feb '13

Release candidate files for 1.53.0 are available at http://boost.cowic.de/rc/ As always, the release managers would appreciate it if you download the candidate of your choice and give building it a try. Please report both success and failure, and anything else that is noteworthy. This helps ensure the candidates build OK before we push them out to SourceForge. The files (and associated md5s) are: 57a9e2047c0f511c4dfcf00eb5eb2fbb *boost_1_53_0_rc1.tar.gz a00d22605d5dbcfb4c9936a9b35bc4c2 *boost_1_53_0_rc1.tar.bz2 c618e030fd4882e4dbacf54baf824544 *boost_1_53_0_rc1.zip cc680cab53a5405ca102a10d43b92b88 *boost_1_53_0_rc1.7z Thanks! -- The release managers

9 10

[container] static_vector: fixed capacity vector update
by Andrew Hundt 01 Feb '13

01 Feb '13

*static_vector is a hybrid of boost::container::vector and boost::array with fixed capacity. Adam Wulkiewicz and I have updated static_vector with improvements from the list discussion and reorganized it for possible inclusion into boost.container. Overview: static_vector is a sequence container with contiguous storage that can change in size like boost::container::vector. It also includes static allocation, low overhead, and fixed capacity similar to boost::array. Synopsis: template* *<* * typename Value, * * std::size_t Capacity, * * typename Strategy = strategy::def<https://svn.boost.org/svn/boost/sandbox/static_vector/doc/html/static_vecto…> <Value> * *>* *class static_vector; Example: // static_vector of ints, fixed capacity: 3 boost::container::static_vector<int,3> three; // size: 0 three.push_back(1); // size: 1 three.push_back(2); // size: 2 three.push_back(3); // size: 3 //three.reserve(4); // no effect, fixed capacity: 3 //three.push_back(3); // size: 4, undefined behavior three.pop_back(); // size: 2 three.shrink_to_fit(); // no effect, fixed capacity: 3 Documentation: https://svn.boost.org/svn/boost/sandbox/static_vector/doc/html/index.html Source: https://svn.boost.org/svn/boost/sandbox/static_vector/ Discussions from boost.devel archive: http://goo.gl/PKEpB [google groups] Changes: - C++11 support - moved to boost::container namespace - strategy based error handling (“strategy” is aka “policy”) - bounds checks are asserts by default but can be switched to exceptions - memory is uninitialized until objects are inserted - internal size is currently the same as Strategy::size_type - expanded documentation and unit tests - optimizations based on type traits - boost::interprocess support ------------------------------------------------ ------------------------------------------------ Major questions from prior list discussions and the related design decisions: ------------------------ Q1: static_vector design *a) combination of fixed capacity vector and adjustable size array b) vector with small size optimization, possibly with configuration of what “small size” means. A1: (a) is implemented by static_vector, leaving the (b) as future work for a separate class. ------------------------ Q2: Is it best to implement push_back() and similar functions with or without bounds check? Main use cases: *a) no bounds check on each insertion because it is an undesirable use of resources b) vector emulation where static_vector is a drop in replacement and optimization so bounds checking is essential to minimize the amount of code that must be changed. A2: (a) no bounds check is the default for static_vector. A vector emulation strategy can be implemented for (b). ------------------------ Q3: Should a failed bounds check trigger an assertion or an exception? Main use cases: a) bounds check is too much overhead, but assert() allows for testing in debug mode b) vector emulation, so bad_alloc should be thrown when the capacity is exceeded c) exceptions are desired when bounds checking, but bad_alloc would cause the wrong behavior because freeing up memory will not allow the possibility of a larger capacity as it would in a vector. A3: (a) failed bounds checks trigger an assertion, and the user can implement the necessary strategy and traits to achieve (b) or (c). ------------------------------------------------ ------------------------------------------------ New/Unanswered Questions: ------------------------ Q4: Should the current static_vector actually be in the detail namespace, with policy specializations providing the actual interface to several classes with different desired functionality? This could be similar to what was suggested by Nate Ridge: On Sat, Oct 15, 2011 at 5:16 PM, Nathan Ridge <zeratul976(a)hotmail.com> wrote: > Now the implementor of these classes can implement them like this: > > typedef boost::detail::static_vector<T, AssertPolicy> capacity_array; > typedef boost::detail::static_vector<T, ThrowPolicy> stack_vector; It also seems to be backed by the fundamental interface differences between the various classes (static_vector, std::vector, AutoBuffer/hybrid_vector) as outlined by Dave Abrahams: On Wed, Oct 19, 2011 at 1:26 PM, Dave Abrahams <dave(a)boostpro.com> wrote: > I see a container that can never store more than N items, for some > reasonably small N, as fundamentally, semantically different from one > that is highly likely to be able to store any M items you have to deal > with, and for which you can't determine a hard limit N ahead of time > (no, max_size() doesn't count—that's typically just > numeric_limits<size_t>::max()). I grant you that we don't have a very > strict way of expressing this difference in a specification, but I think > that's just missing. I wouldn't, except in very special circumstances, > consider one of those containers to be a suitable replacement for the > other, despite the similarity of the specification text. Would you? This would mean that the official interfaces can be specific to various common use cases and a bit simpler. An additional benefit here is that simpler interfaces are more welcoming to a wider range of developers. On the other hand, the current design makes the official interface configurable so they could customize it without being as concerned about changes in the detail implementation. If everyone will want something different for their specific use case, leaving the design as is may make the most sense. Furthermore, a reasonable default provides a simpler interface while permitting advanced configuration for those who need it. ------------------------ Q5: Should the static_vector Strategy concept (a) remain as is (b) add the Capacity as an additional template parameter (c) permit Standard Library Allocators as a strategy Currently, the default traits assume that the Strategy defines types normally defined by Allocator and provides a capacity check failure callback. Also, a different strategy and traits specialization can be used to configure the class as desired, such as modifying the size_type based on the Capacity. Consequently, leaving the Strategy as is (a) remains fairly reasonable. Adding an additional Capacity template parameter (b) to the Strategy concept would take the Strategy further from the Allocator concept, but ensure configuration of size_type based on the capacity is easy and consistent. We believe (c) does not make sense despite allowing static_vector to be more compatible with vector due to the fundamental interface differences discussed in Q4. Option (c) makes more sense for a class like AutoBuffer/hybrid_vector than it does for static_vector. Thus, the question ultimately becomes one concerning the tradeoff between options (a) and (b). ------------------------ Q6: Should static_vector_traits be part of the public interface or remain in the detail namespace? This would make some of the functionality in Q5a part of the official interface, but restrict future changes. In Boost.Container traits are defined in the detail namespace so we were hoping to find out the reasoning for this choice. The direction of the design discussion for Q4 could also make this a moot point. ------------------------ Q7: Does static_vector provide significant benefits over boost::container::vector with a stack Allocator template parameter? This has been discussed previously and we believe the answer is yes for the reasons outlined in Q4 and the documentation, but we wanted to bring it up again in case someone wanted to play devil’s advocate. ------------------------ That covers all the basics of the updated static_vector implementation and some of the biggest outstanding design questions. We appreciate any and all questions, design ideas, or issues you can come up with, so fire away. :-) Regards, Andrew Hundt* Adam Wulkiewicz

7 22

Re: [boost] [variant] Please vote for behavior
by Markus Klein 01 Feb '13

01 Feb '13

On Fri, Jan 25, 2013 at 10:21 AM, Dave Abrahams <dave(a)boostpro.com> wrote: > Yes. IIUC the question here is whether the invariant of variant [;-)] > shall be weakened to accommodate efficient move semantics, thereby > breaking some code, or not, at some cost (the specific costs to be > incurred by various strategies presently under discussion). I'm new to this list, so I won't be offended if you correct me on any rookie mistakes. Concerning the invariant of variant: Why not provide a specialization of boost::optional for variant, which supports move Semantics? Users who don't want their invariant harmed, can use boost::variant as is. Users who need to squeeze out performance could use an optional<variant> and would be explicitly aware of the new null state they introduced for this. Markus www.clean-cpp.org

6 12

Is there any interest in static constructors?
by Alexander Stoyan 01 Feb '13

01 Feb '13

Hello community, I have implemented static constructors for my personal needs. They guarantee construction of static data prior main execution, and allow control the order of static initialization. The code now implemented and tested for VS2012 and GCC. I couldn't find anything like this in the boost library but consider it to be useful for C++ developers. More detailed approach on my blog page: http://alexander-stoyan.blogspot.com/2013/01/static-constructors-in-c.html Feedbacks are very appreciated. Thank you. ______________________________ With best regards, Alexander Stoyan

2 1

Re: [boost] [filesystem] thread safety on msvc 2010
by Jacob Schloss 01 Feb '13

01 Feb '13

> Message: 9 > Date: Mon, 28 Jan 2013 11:05:07 -0500 > From: Beman Dawes <bdawes(a)acm.org> > To: boost(a)lists.boost.org > Subject: Re: [boost] [filesystem] thread safety on msvc 2010 > Message-ID: > < > CAAygHNO8u+gvpGABgJR-KxSP-EqvhyRWRCSKKwx_7chH2hesSg(a)mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > Some critical points to getting a patch accepted are that it applies > only to the affected platform (i.e. Windows) and that it includes test > cases that fail with the current code but pass with the patch applied. > If it works for you with msvc 2010, I can test with other versions and > compilers. > > Thanks, > > --Beman > > Ok, I have some test code I could submit. Is it all right to use boost::threads in the test so it is cross platform? It might be a while (have trip for work soon) until I have time to pull svn (or git, looks like that move is happening soon) and make the patch, but I will try to follow up on this when I get back. Thanks, Jacob PS: Sorry if this broke threading. I just switched off of the digest

2 1