July 2009 - Boost - lists.boost.org

[ann] Urdl - a library for downloading web content
by Christopher Kohlhoff 02 Jul '09

02 Jul '09

Hi all, I have just released a new Boost.Asio-based library, Urdl, which can be used for accessing and downloading web content. <http://think-async.com/Urdl> <http://sourceforge.net/projects/urdl> It currently has limited support for the protocols "http", "https" and "file". It provides an easy-to-use extension to standard C++ iostreams and an asynchronous interface for use with Boost.Asio. In addition to (hopefully) being useful, it is intended as an example of using the Boost.Asio and Boost.System libraries. If you're writing protocol implementations using Asio, you might also be interested in how it uses coroutines to simplify the expression of asynchronous control flow. (Note: here coroutines refers to a macro-based system similar to <http://www.chiark.greenend.org.uk/~sgtatham/coroutines.html>, not Giovanni Deretta's GSoC project). Cheers, Chris

11 20

Fixed point implicity
by Soren Holstebroe 02 Jul '09

02 Jul '09

Hi, I'm generalizing my fixed point class and have come across a few design decisions. 1. Return type of additive arithmetic operation between integers and fixed points Example: a 16 bit integer is added to a 16 bit fixed point with 8 bit fraction. I see three possible return types: 16 bit integer, 16:8 fixed point and promoted 32:8 fixed point If return type is 16 bit integer, the fraction is lost. Clearly not an option. If return type is 16:8 fixed point overflow is possible. A return type of 32:8 would eliminate overflow, but promotion quickly becomes problematic for larger integer types, lets say a 32 bit integer added to a fixed point. In that case we would need a 64 bit integer which could cripple performance on many systems. Another problem with promotion is that the arithmetic expression typically would be in a context where the expression is assigned to a fixed point of the type present in the expression. In that case we have an overflow problem again. I think I favor keeping the return type in 16:8. The user would then have to use an explicit cast to avoid overflow. 2. Return type of additive arithmetic operation between float and fixed points Example: fixed_point<int, 16>(10.5) + 5.5f (integer is representing type, 16 bit is the bit reserved for the fraction) What would you expect the return type to be? I think I favor returning a float. The accuracy of the fixed point is compromised anyway. 3. Return type of additive arithmetic operation between different fixed point types Example: fixed_point<int, 16>(10.5) + fixed_point<int, 8>(10.5) Like in case 1 there is the choice between losing precision, risking overflow or promoting (ex. to fixed_point<boost::int_t<16 + 8>::fast, 16>) I think I favor to simply disallow implicit arithmetic between different fixed point types. This would require the user to cast either type. The problem with that choice is less transparent use of the fixed_point type, since changing a type could break existing code where the previous type was in a valid arithmetic operation with another fixed point type. 4. Similar problems exist for multiplication and division If a fixed_point is multiplied with an integer there is a risk of overflow. If a fixed_point is multiplied with another fixed_point there is a risk of overflow and precision will be lost. If a fixed_point is divided by an integer there is a risk of loss of precision. If a fixed_point or integer is divided by another fixed_point there is a risk of overflow and loss of precision. Soren

5 8

Messages lost
by Chard 02 Jul '09

02 Jul '09

I made two posts to this group on 30 June, around 21:00 UTC. Both received acknowledgements, but neither have appeared. A post to the users group, made at about the same time, appeared there immediately. New messages later than my posting time are appearing, so it doesn't look like it's back-log. Are there any known problems? Apologies if this should be directed elsewhere, but at the moment there doesn't seem to be a route open to www.boost.org (to find out); certainly from my ISP...sigh. (Posting this one at 21:30 UTC)

2 2

[task] 0.2.1 Doesn't compile on VC7.1 / older gcc
by Chard 02 Jul '09

02 Jul '09

Hi Oli, As documented, you're testing MS compilation with VC9. We're currently tied to VC7.1 and as we use a threadpool heavily (Philipp Henkel's) I have a particular interest in seeing an 'official' boost threadpool. So forgive me if I keep these coming ;-) There is another VC7.1 compiler conformance issue: static_pool now has the contained pool_base class. The pool_base::submit method constructs a handle<>, however, this has the VC7.1 'friend access to contained class' problem. The handle<> (non-default) constructor is private, granting friend access to static_pool, however, static_pool<>::pool_base requires access, and VC7.1 barfs. I couldn't figure out the friend syntax (that works) to add to class handle<>, so for now I'm working with handle<> (hacked as) public. (Note: earlier versions of gcc, e.g. 3.4.4, will exhibit this problem, too.) I thought this would've been the syntax: # if BOOST_WORKAROUND(BOOST_MSVC, <= 1300) template <typename Channel> friend class static_pool<Channel>::pool_base; # endif but it doesn't appear to be legal. Is this type of declaration even possible? An alternative would be to move (some or all of) the functionality to the outer static_pool class, and route the inner pool_base class to it, e.g. template< typename R > handle< R > static_pool::pool_base::submit( boost::detail::thread_move_t< task< R > > t_) { static_pool::submit(*this, _t); } template< typename R > static handle< R> static_pool::submit(pool_base &pb_, boost::detail::thread_move_t< task< R > > t_) { ... } Or, perhaps just the handle creation line. However, it may be tidier just to have a detail::create_handle (that is the handle<> friend), seeing as you are just trying to say nobody else should be constructing a handle. Regards, and keep up the good work...

2 1

[task] 0.1 => 0.2.1 removes make_task
by Chard 02 Jul '09

02 Jul '09

Task version 0.1 => 0.2.1 comment. make_task appears to have been removed and the multiple argument mechanism migrated to the task constructor. AFAICT this has removed functionality because the make_task helper had the benefit of determining the return value of the task. The task class is templated on the return type, so the user if forced to declare it (okay, not a great deal, but then we have conveniences like make_pair). e.g. 0.1: task::async(threadpool, task::make_task(TaskFunc)); 0.2.1 (argument order exchanged in 0.2.1): task::async(move(task::task<void>(TaskFunc)), threadpool); I assume the reasoning behind this was the cut-over to using move semantics; if this is the case then the constructor: explicit task( Fn fn, BOOST_ENUM_TASK_MAKE_TASK_FUNC_ARGS(n)) \ : task_( new detail::task_wrapper< \ typename result_of< Fn( BOOST_PP_ENUM_PARAMS(n, A)) >::type, \ function< typename result_of< Fn( BOOST_PP_ENUM_PARAMS(n, A)) >::type() > \ >( bind( fn, BOOST_PP_ENUM_PARAMS(n, a)) ) ) \ {} may as well not use result_of; is this not just the template argument, R ? Could make_task have been preserved, but modified to return the move-type ?

2 1

boost::interprocess - lnk2005 error due to compiler/linker bug in msvc8
by willem hengeveld 02 Jul '09

02 Jul '09

when using boost::interprocess under win32, i ran into the following problem: in boost/interprocess/detail/win32_api.hpp several static inline functions have a function local class for releasing resources, like 'unlink_file', which defines 2 classes: handle_closer, and auto_ptr due to a compiler bug, as discussed here: http://www.eggheadcafe.com/conversation.aspx?messageid=32706692&threadid=32… you get a LNK2005 error <symbol> already defined in <object> when using boost::interprocess from multiple .cpp files a solution is to use boost::shared_ptr instead of local classes in win32_api.hpp, like this: boost::shared_ptr<void> hiPSAPI = boost::shared_ptr<void>(load_library("PSAPI.DLL"), free_library); attached is the complete patch willem

3 2

[swap] please update documentation with the header I need to include
by Thorsten Ottosen 01 Jul '09

01 Jul '09

-Thorsten

8 20

Re: [boost] Fixed point integer proposal
by Soren Holstebroe 01 Jul '09

01 Jul '09

> > Some years ago I wrote a fixed point template for transparent fixed point > > integer arithmetics and conversion. > > To me, "fixed point" => !"integer". Obviously fixed point is not an integer in mathematical sense. What I meant was that a fixed point represented by an integer shares some of the same computational properties: Arithmetic is faster than floats on most architectures. Adding (and accumulating) fixed points or multiplying by an integer is exact. Conversion to integer is fast. > particular, saturating or otherwise detecting overflows seemed to be > required. Saturation and overflow detection can be specified as a policy. For me there are two important design criteria: 1st priority is that computational performance matches integer performance. 2nd priority is that it is possible to make a seamless float type and the fixed point type exchange. Adding saturation handling and overflow checks, that are not natively supported by the underlying integer is of lower priority for me. In my fixed point class I implemented overflow check as an optional policy. That way you can add all the overhead checks you want in your debug configuration and strip it in your release build. As of saturation (I assume it's computer graphics people that want that feature) it should be straight forward to implement that as a policy too in the same way as the overflow check. In my option I see little practical use of a saturation policy. Saturation code would kill performance and people would be better off using MMS registers for native saturation. > recall that someone was actually working on that. It was also > suggested that I ought to provide all of the features that <cmath> > provides for floating point. It makes sense to support rounding, abs and mod. I don't see a practical need to natively support trigonometry functions. A conversion to float and back has to be fine. Soren

5 10

[config] problem with sys/select.h on Linux/Unix
by Iain K. Hanson 01 Jul '09

01 Jul '09

Hi, boost config seems to be indirectly including sys/select.h on Linux. This causing me a problem in my sockets library where I have my own declarations of select, fd_set etc which cause duplicate symbols errors. No only can I not use config but also any boost lib as they all use config :-(. Can someone please suggest a fix for this? Out of curiosity, why does config use select.h? regards /ikh -- "I disapprove of what you say, but I will defend to the death your right to say it," Evelyn Beatrice Hall who wrote under the pseudonym S.G. Tallentyre in a biography of Voltaire. #include <std_disclaimer>

2 4

[serialization] Memoryleak in void_caster::recursive_register
by Runar Undheim 01 Jul '09

01 Jul '09

// implementation of void caster base class BOOST_SERIALIZATION_DECL(void) void_caster::recursive_register(bool includes_virtual_base) const { void_cast_detail::set_type & s = void_cast_detail::void_caster_registry::get_mutable_instance(); s.insert(this); // generate all implied void_casts. void_cast_detail::set_type::const_iterator it; for(it = s.begin(); it != s.end(); ++it){ if(m_derived == (*it)->m_base) new void_caster_shortcut( (*it)->m_derived, m_base, m_difference + (*it)->m_difference, includes_virtual_base ); if((*it)->m_derived == m_base) new void_caster_shortcut( m_derived, (*it)->m_base, m_difference + (*it)->m_difference, includes_virtual_base ); } } The new void_caster_shortcut in void_caster::recursive_register give memory leak. The leak is only at startup of program, so it doesn't eat any memory while the program is running. But it makes it difficult to find other memory leaks when the program contain several leaks you don't care about. - Runar

2 2