- Boost-users - lists.boost.org

Rational: [PATCH] Eliminate excessive calls to gcd and add operator%
by Mario Lang 22 May '12

22 May '12

Hi. I am using boost/rational.hpp in my code to represent musical time with rational numbers. I have found some possible improvements. These improvements have accumulated into a small patch which I'd like to present for review. I think it is worth applying. There doesn't appear to be a maintainer for rational.hpp though, so I am writing to the list. The story: In my application I need a remainder operation. Looking at rational.hpp and its work-horse operators.hpp it became rather obvious that rational.hpp can and (in my opinion) should implement operator%= and friends. operator%= can easily be defined as return operator-= (other * rational_cast<int_type>(*this / other)); I have been using this definition for operator%= successfully in my code for quite a while. During a profiling session it showed up which made me think a bit more. Actually, the current way how rational.hpp implements mixed-mode operators is a waste of performance. It calls the corresponding operator with a temporary rational where the denominator is always equal to 1. However, this means that there are excessive (unnecessary) calls to boost::math::gcd performed by the operators. For instance, addition of two rational numbers needs two calls to gcd. On the other hand, addition of a rational number and an integer can be performed without any gcd at all (n += d * i). Similarily, multiplication of two rational numbers needs two calls to gcd, while multiplication of a rational with an integer only needs one call to gcd. So it is definitely worth it to actually define the mixed mode operators not in terms of the already existing operators but reimplement them with the unnecessary calls to gcd removed. After this, the definition of operator%=(const rational&) results in one less call to gcd (because there is one mixed-mode multiplication involved) and the definition of operator%=(param_type) saves two unnecessary calls to gcd since there is one mixed-mode multiplication and one mixed-mode division involved. Obviously, all code that involves mixed-mode operators benefits from these changes. As a side effect of reading code I noticed that the definition of operator=(param_type) (assign from an integer) also does unnecessary gcd. It currently calls assign(n, 1) which itself calls normalize() which does gcd. However, as above, we know that there is no need to calculate gcd of (n, 1) since it will always be 1. So I changed the definition of operator=(param_type) to directly assign the parameter to num and set den to 1, instead of calling assign(). Another useless call of gcd squashed! On top of these, we can observe that now that boost::rational provides operator%= it is actually an instance of ordered_euclidean_ring_operators. Comments and committers welcome. Index: rational.hpp =================================================================== --- rational.hpp (Revision 78523) +++ rational.hpp (Arbeitskopie) @@ -21,6 +21,7 @@ // Nickolay Mladenov, for the implementation of operator+= // Revision History +// 05 May 12 Add operator% and reduced use of implicit gcd (Mario Lang) // 05 Nov 06 Change rational_cast to not depend on division between different // types (Daryle Walker) // 04 Nov 06 Off-load GCD and LCM to Boost.Math; add some invariant checks; @@ -105,23 +106,10 @@ template <typename IntType> class rational : - less_than_comparable < rational<IntType>, - equality_comparable < rational<IntType>, - less_than_comparable2 < rational<IntType>, IntType, - equality_comparable2 < rational<IntType>, IntType, - addable < rational<IntType>, - subtractable < rational<IntType>, - multipliable < rational<IntType>, - dividable < rational<IntType>, - addable2 < rational<IntType>, IntType, - subtractable2 < rational<IntType>, IntType, - subtractable2_left < rational<IntType>, IntType, - multipliable2 < rational<IntType>, IntType, - dividable2 < rational<IntType>, IntType, - dividable2_left < rational<IntType>, IntType, - incrementable < rational<IntType>, - decrementable < rational<IntType> - > > > > > > > > > > > > > > > > + ordered_euclidean_ring_operators < rational<IntType>, + ordered_euclidean_ring_operators2 < rational<IntType>, IntType, + unit_steppable < rational<IntType> + > > > { // Class-wide pre-conditions BOOST_STATIC_ASSERT( ::std::numeric_limits<IntType>::is_specialized ); @@ -141,7 +129,7 @@ // Default copy constructor and assignment are fine // Add assignment from IntType - rational& operator=(param_type n) { return assign(n, 1); } + rational& operator=(param_type i) { num = i; den = 1; return *this; } // Assign in place rational& assign(param_type n, param_type d); @@ -155,15 +143,17 @@ rational& operator-= (const rational& r); rational& operator*= (const rational& r); rational& operator/= (const rational& r); + rational& operator%= (const rational& r); - rational& operator+= (param_type i); - rational& operator-= (param_type i); + rational& operator+= (param_type i) { num += i * den; return *this; } + rational& operator-= (param_type i) { num -= i * den; return *this; } rational& operator*= (param_type i); rational& operator/= (param_type i); + rational& operator%= (param_type i); // Increment and decrement - const rational& operator++(); - const rational& operator--(); + const rational& operator++() { num += den; return *this; } + const rational& operator--() { num -= den; return *this; } // Operator not bool operator!() const { return !num; } @@ -205,6 +195,14 @@ void normalize(); }; +// Type conversion +template <typename T, typename IntType> +inline T +rational_cast(const rational<IntType>& r BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(T)) +{ + return static_cast<T>(r.numerator()) / static_cast<T>(r.denominator()); +} + // Assign in place template <typename IntType> inline rational<IntType>& rational<IntType>::assign(param_type n, param_type d) @@ -327,50 +325,54 @@ return *this; } -// Mixed-mode operators template <typename IntType> inline rational<IntType>& -rational<IntType>::operator+= (param_type i) +rational<IntType>::operator%= (const rational<IntType>& r) { - return operator+= (rational<IntType>(i)); + return operator-= (r * rational_cast<IntType>(*this / r)); } +// Mixed-mode operators template <typename IntType> inline rational<IntType>& -rational<IntType>::operator-= (param_type i) -{ - return operator-= (rational<IntType>(i)); -} - -template <typename IntType> -inline rational<IntType>& rational<IntType>::operator*= (param_type i) { - return operator*= (rational<IntType>(i)); + // Avoid overflow and preserve normalization + IntType gcd = math::gcd(i, den); + num *= i / gcd; + den /= gcd; + + return *this; } template <typename IntType> -inline rational<IntType>& +rational<IntType>& rational<IntType>::operator/= (param_type i) { - return operator/= (rational<IntType>(i)); -} + // Avoid repeated construction + IntType const zero(0); -// Increment and decrement -template <typename IntType> -inline const rational<IntType>& rational<IntType>::operator++() -{ - // This can never denormalise the fraction - num += den; + if (i == zero) throw bad_rational(); + if (num == zero) return *this; + + // Avoid overflow and preserve normalization + IntType const gcd = math::gcd(num, i); + num /= gcd; + den *= i / gcd; + + if (den < zero) { + num = -num; + den = -den; + } + return *this; } template <typename IntType> -inline const rational<IntType>& rational<IntType>::operator--() +inline rational<IntType>& +rational<IntType>::operator%= (param_type i) { - // This can never denormalise the fraction - num -= den; - return *this; + return operator-= (i * rational_cast<IntType>(*this / i)); } // Comparison operators @@ -477,12 +479,7 @@ template <typename IntType> bool rational<IntType>::operator> (param_type i) const { - // Trap equality first - if (num == i && den == IntType(1)) - return false; - - // Otherwise, we can use operator< - return !operator<(i); + return operator==(i)? false: !operator<(i); } template <typename IntType> @@ -583,24 +580,13 @@ return os; } -// Type conversion -template <typename T, typename IntType> -inline T rational_cast( - const rational<IntType>& src BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(T)) -{ - return static_cast<T>(src.numerator())/static_cast<T>(src.denominator()); -} - // Do not use any abs() defined on IntType - it isn't worth it, given the // difficulties involved (Koenig lookup required, there may not *be* an abs() // defined, etc etc). template <typename IntType> inline rational<IntType> abs(const rational<IntType>& r) { - if (r.numerator() >= IntType(0)) - return r; - - return rational<IntType>(-r.numerator(), r.denominator()); + return r.numerator() >= IntType(0)? r: -r; } } // namespace boost -- CYa, ⡍⠁⠗⠊⠕

2 3

Problem compiling with boost thread
by Rajalakshmi Iyer 21 May '12

21 May '12

Hello, I am trying to compile my application with Boost thread library (1.49.0) on Ubuntu 11.10 (gcc 4.6.1). However I see this error: $(BOOST_PATH)/include/boost/thread/detail/config.hpp:29:24: fatal error: platform.hpp: No such file or directory compilation terminated. In the file $(BOOST_PATH)/include/boost/thread/detail/config.hpp, the platform.hpp is included as: #include "platform.hpp" If I change that to: #include <boost/thread/detail/platform.hpp> this error goes away. However, there are a lot of places where headers are included using relative path, thereby causing compilation errors. Is there a workaround for this? Thanks, Raj -- Twitter: @Blismedia <http://twitter.com/#%21/blismobile> BlisMedia 32 Percy Street, London W1T 2DE www.blismedia.com <http://www.blismobile.com/> <http://www.blismobile.com/> [image: Follow on Twitter] <http://twitter.com/#%21/blismobile>[image: Blis Website] <http://www.blismobile.com/> This communication is from BlisMedia, which is a trading name of Breeze Tech (UK) Ltd, a company registered in England and Wales with registered number 06455773. Its registered office is 32 Percy Street, London W1T 2DE, United Kingdom. This communication contains information that is confidential and may also be privileged. It is for the exclusive use of the intended recipient(s). If you are not the intended recipient(s), please (1) notify info(a)blismedia.com<info(a)blismobile.com> by forwarding this email and delete all copies from your system and (2) note that disclosure, distribution, copying or use of this communication is strictly prohibited. Email communications cannot be guaranteed to be secure or free from error or viruses. All emails sent to or from a Blismobile email account are securely archived and stored by an external supplier. This email does not constitute a contractual agreement; such agreements are in specified contractual or Insertion Order (IO) form only and exclusively contain all the terms to which Breeze Tech )UK) Ltd will be bound. To the extent permitted by law, Breeze Tech (UK) Ltd does not accept any liability for use of or reliance on the contents of this email by any person save by the intended recipient(s) to the extent agreed in a contract or Insertion Order. Opinions, conclusions and other information in this email which have not been delivered by way of the business of Breeze Tech (UK) Ltd are neither given nor endorsed by it.

2 1

Re: [Boost-users] asio - improving chat server example for my
by Marat Abrarov 21 May '12

21 May '12

Hi, John. > Many thanks for any help. I'm really enjoying Asio and other parts of Boost, but I still don't > understand the framework completely. Sorry, if I'm spamming you but feel free to read Asio examples at my project - http://sourceforge.net/projects/asio-samples. May be it can help in some architecture decisions. > 2- This might be way off, but I also wondered if I need to close all the sockets so that the thread group could join. Yes, you need to close all sockets (and other instances of asio::basic_io_object) having pending asynchronous operations (read/write for sockets, async_wait for timers, etc), related to the used instance of asio::io_service. Otherwise you can call asio::io_service::stop(). > I added 2 member functions to help me: > > void chat_server::close() > { > ? room_.close(); > } > > void chat_room::close() > { > ? std::for_each( participants_.begin(), participants_.end(), boost::bind(&chat_room::leave, this, _1) > ); > } > > I call this in my server app class's destructor before calling thread_group::join_all, but this has a > crash. The crash occurs right after chat_room::close() completes. I'm not confident in the way I used > bind(...), but I was able to see the function execute, and then the crash occurs later. You need to implement chat_participant::close() like it is done at: http://www.boost.org/doc/libs/1_49_0/doc/html/boost_asio/example/chat/chat_… Regards, Marat Abrarov.

2 1

Deriving custom archive classes from boost::archive::text_oarchive_impl and boost::archive::text_iarchive_impl
by Aldo Longhi 21 May '12

21 May '12

*Note: Boost's archive scheme is based on symmetrical input and output archive classes. It's tedious to write about both of them all the time, so I'll use "?archive" to mean both "oarchive" and "iarchive".* *Summary:* After changing the base classes of my custom archives from binary_?archive_impl to text_?archive_impl, my custom archive classes are no longer "found" when the compiler is instantiating the serialize(...) methods in my other classes. *Background:* My application was successfully reading and writing files to disk using subclasses of binary_?archive_impl (the documentation and/or code comments recommend this over deriving from binary_oarchive). I needed to switch from a binary file format to a text format, so I switched the base classes of my custom archives to text_?archive_impl. That's when everything blew up. *The problem:* My custom archive classes add functionality, including some additional methods which do not exist in their Boost base classes; these methods are called in the serialize(...) methods in many of my classes, and they were working fine. After changing the base classes from binary_?archive_impl to text_?archive_impl, I received compilation errors all over the place complaining that my custom methods do not exist in text_?archive. Well, that's obvious (!!!), but they DO exist in my CUSTOM archives, and they were working JUST FINE when I was using Boost's binary base classes. What's the deal? *What I found, and my temporary - but undesirable - solution:* After tearing my hair out and going around in circles for about a day, this is what I found... 1) Some time ago (Boost 1.34 I believe), the files "binary_?archive.hpp" were split up into "binary_?archive_impl.hpp" and "binary_?archive.hpp" (the latter #include the former). This WAS NOT done to "text_?archive.hpp". (As a result, I changed my application's #include lines from "binary_?archive_impl.hpp" to simply "text_?archive.hpp".) 2) If I split up "text_?archive.hpp" into two parts and #include only the "..._impl.hpp" headers, everything works. (But I really DON'T want to modify my Boost installation!) 3) Looking more closely at these headers and fiddling around a bit, I found that if I use the original, unmodified headers and comment out the line BOOST_SERIALIZATION_REGISTER_ARCHIVE(boost::archive::text_oarchive) (and likewise for text_iarchive), then everything works fine again. (By the way I have similar lines in my own archive code to "register" my custom archives.) *The mystery, and my dilemma:* A) Why does the presence of those lines foul up the works? ...and why does removing them make things work? ...and what might I have broken (without knowing it) by doing so? B) Why were the "text_?archive.hpp" files NOT split up along with the "binary_?archive.hpp" files a long time ago? (Is the library broken? Should it be fixed?) C) Is there any way to solve this in my application code WITHOUT modifying my Boost installation? *P.S. I'm using Boost 1.48 and Visual Studio 2010 SP1 (64-bit)*

1 0

Re: [Boost-users] Status of boost.lockfree
by Juan Ramírez 21 May '12

21 May '12

Why don't you try on boost-dev mailing list? Saludos! Juan On Mon, May 21, 2012 at 12:59 PM, Juan Ramírez <jramirez(a)inconcertcc.com> wrote: > Why don't you try on boost-dev mailing list? > > > Saludos! > Juan > > > > > On Sat, May 19, 2012 at 8:00 AM, niXman <i.nixman(a)gmail.com> wrote: >> >> There is someone alive? >> Or, I asked a very difficult question? >> >> Thanks! >> >> -- >> Regards, >> niXman >> _______________________________________________ >> Boost-users mailing list >> Boost-users(a)lists.boost.org >> http://lists.boost.org/mailman/listinfo.cgi/boost-users > >

1 0

Compiler warns about 'reference to temporary' with iterator on shared memory map
by Rajalakshmi Iyer 21 May '12

21 May '12

Hello, Consider: typedef std::pair<int, int> value_type; typedef allocator<value_type, managed_shared_memory::segment_manager> ValueAllocator; typedef boost::interprocess::map<int, int, std::less<int>, ValueAllocator> A_map; class A { public: A() { /* Construct shared memory map */ } ~A() bool contains(int i); private: A_map* m_objA; }; inline bool A::contains(int i) { A_map::iterator itr = m_objA->find(i); return (itr != m_objA->end()); } Here, at the line A_map::iterator itr = m_objA->find(i); I see this warning: .../boost_1_49_0/include/boost/container/detail/tree.hpp:391:31: warning: returning reference to temporary [enabled by default] I dont see a reason for this. Can someone please point out if I am doing something wrong here? Thanks in advance, Regards, Raj -- Twitter: @Blismobile <http://twitter.com/#!/blismobile> BlisMobile Media 32 Percy Street, London W1T 2DE www.blismobile.com [image: BlisMobile] <http://www.blismobile.com/>[image: Follow on Twitter]<http://twitter.com/#!/blismobile>[image: Blis Website] <http://www.blismobile.com/> This communication is from BlisMobile Media, which is a trading name of Breeze Tech (UK) Ltd, a company registered in England and Wales with registered number 06455773. Its registered office is 32 Percy Street, London W1T 2DE, United Kingdom. This communication contains information that is confidential and may also be privileged. It is for the exclusive use of the intended recipient(s). If you are not the intended recipient(s), please (1) notify info(a)blismobile.com by forwarding this email and delete all copies from your system and (2) note that disclosure, distribution, copying or use of this communication is strictly prohibited. Email communications cannot be guaranteed to be secure or free from error or viruses. All emails sent to or from a Blismobile email account are securely archived and stored by an external supplier. This email does not constitute a contractual agreement; such agreements are in specified contractual or Insertion Order (IO) form only and exclusively contain all the terms to which Breeze Tech )UK) Ltd will be bound. To the extent permitted by law, Breeze Tech (UK) Ltd does not accept any liability for use of or reliance on the contents of this email by any person save by the intended recipient(s) to the extent agreed in a contract or Insertion Order. Opinions, conclusions and other information in this email which have not been delivered by way of the business of Breeze Tech (UK) Ltd are neither given nor endorsed by it.

3 6

[asio] Correct handler signature for io_service::post/dispatch
by Gary Sanders 20 May '12

20 May '12

Currently I'm developing a server that uses Boost.Asio, version 1.49. It makes use of io_service::post and io_service::dispatch to invoke handlers running in the same process, but on different threads. Some of the handlers invoke a function with the following signature: void handler(std::string s); This works fine and no problems have been observed. However, subsequently I read in the asio documentation that the function signature of a handler invoked by post/dispatch *must* be in the following format: void handler(); Is my code working by chance or is the asio version 1.49 documentation out of date? Regards Gary Sanders

3 4

boost asio http server vs indy httpserver component
by Christer Borgqvist 20 May '12

20 May '12

Hi, I have an application writen in Borland C++ Builder, using Indy HTTPServer component to handle each connection in blocking mode, each connection starts a new thread, if I want to do the same using boost what is the appropriate way to do that? appreciate any help. Christer

2 1

[serialization][1.49]How to serialize shared_ptr<void>?
by harr999y 19 May '12

19 May '12

Hi,:) I'm doing saving the shared_ptr<void>,but it can't as "boost::serialization::tracking_level< T >::value != boost::serialization::track_never". And I know if it's shared_ptr<int>,I can use strong_typedef.But for the void,the strong_typedef cannot work. So can you give me some hint? Thanks, Harry -- View this message in context: http://boost.2283326.n4.nabble.com/serialization-1-49-How-to-serialize-shar… Sent from the Boost - Users mailing list archive at Nabble.com.

2 3

Re: [Boost-users] Status of boost.lockfree
by niXman 19 May '12

19 May '12

There is someone alive? Or, I asked a very difficult question? Thanks! -- Regards, niXman

2 1