June 2004 - Boost - lists.boost.org

Request for reviewers for a CUJ article
by Joaquín Mª López Muñoz 27 Jun '04

27 Jun '04

I've just finished an article for publication in CUJ about Boost.MultiIndex (they accepted the preliminary proposal.) The author guidelines suggest that a couple of technical peers do a review of the text before submission (specifically, the purpose is that the peers "can give you candid feedback regarding accuracy and adherence to standard industry practices".) So, I kindly request some volunteer wanting to do the job. The review shouldn't be too time-consuming (about 3,000 words of text.) I guess someone without prior exposure to the library is a better candidate, so as to probe the introductory quality of the article. Due to (c) issues I guess I cannot just put the article on the web for you to download. If you are interested, please contact me privately and I'll send you the text. Thanks very much for your help. Regards, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 2

Re:BOOST_STATIC_ASSERT(false) - was (no subject)
by Robert Ramey 27 Jun '04

27 Jun '04

I hadn't considered Michaels idea. I've been using BOOST_STATIC_ASSERT(0 == sizeof(T)). Michiel Salters wrote: >Robert Ramey wrote: >> BOOST_STATIC_ASSERT(false) >This *should* work. IMHO, if it doesn't, we should fix it, not work >around it like Michiel suggested. I don't think I'ts fixable. But maybe one might want to make another macro named something like BOOST_BETTER_NOT_ARRIVE_HERE - lol >> However, its doesn't work >> with compilers that are too smart for their own good. In order to make this >> work I need something like: >Which compiler and which version of boost and the Compiler are you >using? Can you provide a minimal but complete example? Also, you might >want to have a look at checked_delete where we had similar problems in >the past IIRC. I've noticed it at least on Comeau. I just used the fix above so now I don't see it. So now I don't know which compilers complain. Robert Ramey

2 1

newbie question: Cannot open include file: 'boost/filesystem/operations.hpp': No such file or directory
by Chris 27 Jun '04

27 Jun '04

Hi to all.. sorry for the newbie question. I believe I have built the Boost libs correctly. I used: bjam -s"TOOLS=vc7.1" -s"VC71_ROOT=C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7" install The Includes and Libs are in C:\Boost as described in the "Getting Started Guide" Also I didn't get any errors during the building process (that's why I assume the build was succesfull), so I decided to test the libraries and just took one of the examples form the site. Compiled it and got the: "fatal error C1083: Cannot open include file: 'boost/filesystem/operations.hpp': No such file or directory" #include "boost/filesystem/operations.hpp" #include "boost/filesystem/path.hpp" #include <iostream> namespace fs = boost::filesystem; int main( int argc, char* argv[] ) { //removed for brevity... } I also added the path to the environmental variables in windows, under INCLUDE and LIB, but this didn't seem to change anything. Can anyone help me understand why the compiler doesn't find the headers? What am I missing? Thank you in advance for your help... Chris PS. I am really sorry for the newbie question....

2 1

Problems with conditions in boost::thread
by Andreas Pokorny 27 Jun '04

27 Jun '04

Hi, I just started to use boost::thread in a bigger application. Since the condition variables do not work as expected - e.g. the waiting thread never wakes up ( example can be found here: http://svn.zynot.org/svn/zynot/projects/xeta/trunk/source/src/xeta/curl.cpp ) - I started writing a smaller test case to see if I use the library correctly. That little test application works fine if I run it in with the gnu debugger, and if I have debugging symbols in my glibc. In all other cases the process gets terminated before any debuging print outs show up or the debugger fails to run the application: #include <list> #include <iostream> #include <boost/bind.hpp> #include <boost/thread/xtime.hpp> #include <boost/thread/thread.hpp> using namespace boost; struct Mini { bool running; thread worker; mutex list_mutex; condition items_empty; std::list<std::string> items; typedef mutex::scoped_lock lock; Mini(); ~Mini(); void run(); void add( std::string const& item); }; Mini::Mini() : worker( bind( &Mini::run, this ) ) { } Mini::~Mini() { running = false; { lock the_list( list_mutex ); items_empty.notify_one(); } worker.join(); } void Mini::run() { running = true; while( running ) { lock the_list( list_mutex ); if( running && items.empty() ) { std::cout << "Waiting for item" << std::endl; items_empty.wait( the_list ); std::cout << "Continue" << std::endl; } std::cout << "Removing :" << items.front() << std::endl; items.pop_front(); } } void Mini::add( std::string const& item ) { lock list_access( list_mutex ); std::cout << "Adding :" << item << std::endl; items.push_back( item ); std::cout << "Notifying other thread" << std::endl; items_empty.notify_one(); } int main() { Mini my_mini; my_mini.add("Once upon a time"); my_mini.add("there was a little mouse"); my_mini.add("that lived in a little house"); xtime xt; xtime_get(&xt, TIME_UTC); xt.sec += 2; thread::sleep(xt); } After that I tried to do it without boost. This time everything works as expected, with and without debugger. #include <list> #include <unistd.h> #include <iostream> #include <pthread.h> struct Mini { bool running; pthread_t worker; pthread_mutex_t mutex; pthread_cond_t not_empty; std::list<std::string> items; Mini(); ~Mini(); static void * call_run( void *); void run(); void add( std::string const& item); }; void* Mini::call_run( void * ptr) { static_cast<Mini*>(ptr)->run(); return 0; } Mini::Mini() { pthread_mutex_t temp_mut = PTHREAD_MUTEX_INITIALIZER; mutex = temp_mut; pthread_cond_t temp_cond = PTHREAD_COND_INITIALIZER; not_empty = temp_cond; worker = pthread_create( &worker, 0, Mini::call_run, this ); } Mini::~Mini() { running = false; { pthread_mutex_lock( &mutex ); pthread_cond_broadcast( &not_empty ); pthread_mutex_unlock( &mutex ); } pthread_join( worker, 0 ); pthread_cond_destroy( &not_empty ); } void Mini::run() { running = true; while( running ) { pthread_mutex_lock( & mutex ); if( running && items.empty() ) { std::cout << "Waiting for item" << std::endl; pthread_cond_wait(&not_empty, &mutex); std::cout << "Continue" << std::endl; } std::cout << "Removing :" << items.front() << std::endl; items.pop_front(); pthread_mutex_unlock( &mutex ); } } void Mini::add( std::string const& item ) { pthread_mutex_lock( & mutex ); std::cout << "Adding :" << item << std::endl; items.push_back( item ); std::cout << "Notifying other thread" << std::endl; pthread_cond_broadcast(& not_empty); pthread_mutex_unlock( &mutex ); } int main() { Mini my_mini; my_mini.add("Once upon a time"); my_mini.add("there was a little mouse"); my_mini.add("that lived in a little house"); sleep(2 ); } Any ideas?

2 2

Autolinking problem on Intel (Win32) with STLport
by Doug Gregor 26 Jun '04

26 Jun '04

It looks like there is a problem with autolinking as mentioned in the subject. It looks like a library naming issue: http://tinyurl.com/2djmv Doug

2 2

How to use boost::thread::condition
by Andreas Pokorny 26 Jun '04

26 Jun '04

Hi, I have problems understanding how mutexes, scoped locks and conditions should work togehter. I took the condition.cpp example and with the buffered shared by two threads as a basis of my code: http://svn.zynot.org/svn/zynot/projects/xeta/trunk/source/src/xeta/curl.cpp There is a list filled by one thread, and a thread that removes the entries after they have been downloaded using curl. The worker-thread waits for a jobs_empty condition and the insertion method calls notify_one after every insertion. Currently the worker-thread just prints "Sitting and waiting", and never reacts on any calls of notifiy_one(). Regards Andreas Pokorny

1 1

BOOST_WORKAROUND in test files?
by Gennaro Prota 26 Jun '04

26 Jun '04

Hi, is it allowed/OK to use BOOST_WORKAROUND in the boost test files? Or should we simply use #if/#ifdef? For instance, Codewarrior 8.3 for Windows has some problems in parsing a function template in dyn_bitset_unit_tests4.cpp. A simple fix, suggested by Howard, is: #ifdef __MWERKS__ // will be: __MWERKS__ && __MWERKS__ <= xxx #pragma parse_func_templ off #endif Should this use BOOST_WORKAROUND? In general: What should compiler/library vendors see when they define BOOST_STRICT_CONFIG? Just errors in the boost headers or errors in the test files too? -- Genny.

2 1

intrusive_ptr: proposal for common base class
by Stefan Slapeta 26 Jun '04

26 Jun '04

Hi all, Sorry if this has been discussed yet, but I think there is an obvious pattern in the usage of intrusive_ptr which should be exposed as common functionality in boost. In most of the cases you just want to handle a simple reference count and delete the object which is held by the intrusive_ptr when the reference count is decreased to zero. It's very annoying that you currently have to write external add_ref/release functions for every class you want to use with this smart pointer. Additionaly, it makes the code unreadable. That's why I would like to suggest having a base class which provides the simplest functionality a user could expect from intrusive_ptr: handling the reference count and deleting the object. A class which should be maintained by intrusive_ptr can derive from it and would not have to provide any further interfaces. I'm not glad about the class name 'intrusive_ptr_base' but I don't have any good ideas at the moment :( Awaiting you comments! (code: see below) Stefan namespace boost { class intrusive_ptr_base; void intrusive_ptr_add_ref(intrusive_ptr_base*); void intrusive_ptr_release(intrusive_ptr_base*); class intrusive_ptr_base { friend void intrusive_ptr_add_ref(intrusive_ptr_base*); friend void intrusive_ptr_release(intrusive_ptr_base*); int ref_count__; protected: intrusive_ptr_base() : ref_count__(0) {} virtual ~intrusive_ptr_base() {} }; inline void intrusive_ptr_add_ref(intrusive_ptr_base* o) { ++o->ref_count__; } inline void intrusive_ptr_release(intrusive_ptr_base* o) { if (--o->ref_count__ == 0) { delete o; } } } // namespace

2 2

Hmmm.. how do I add a new test to the regressions?
by Fernando Cacciola 26 Jun '04

26 Jun '04

I forgot it.... I need to add the Numeric Conversions Library so I can start hunting compiler problems. Fernando Cacciola SciSoft

2 1

RE: Boost.Threads proposed change: lock constructors
by Batov, Vladimir 26 Jun '04

26 Jun '04

Mike, With all due respect I am not sure that the additional complexity you are proposing is justified. It appears that the differences in lock and try_lock default constructors' behavior are consistent with other libraries and actually feel intuitive and natural to me. After all, it is reflected in what I do -- with lock I try until I succeed (that is [Batov, Vladimir] the lock blocks until succeeds), with try_lock I try once and may fail (that is it unconditionally falls through for me to check the result). Best, V. Date: Thu, 24 Jun 2004 14:14:02 -0400 From: Michael Glassford <glassfordm(a)hotmail.com> Subject: [boost] Boost.Threads proposed change: lock constructors To: boost(a)lists.boost.org Message-ID: <cbf5as$hvv$1(a)sea.gmane.org> Content-Type: text/plain; charset=us-ascii; format=flowed Below I detail proposed changes to the Boost.Threads lock class constructors. Comments? Questions? Mike ----------------------------------------------------------------- The Boost.Threads lock concept constructors are currently defined as follows: Definitions ----------- L: lock type l: lock variable (type L) m: mutex variable (appropriate Mutex type) s: initial lock state (type bool, true=locked, false=unlocked) Lock Concept ------------ L l(m,s) //Blocking lock if s L l(m) //Blocking lock TryLock Concept --------------- L l(m,s) //Blocking lock if s L l(m) //Non-blocking lock TimedLock Concept ----------------- L l(m,s) //Blocking lock if s L l(m,t) //Blocking lock, failing if not obtained by time t These definitions have the following drawbacks: 1) That one of the TryLock constructors blocks while the other doesn't is neither obvious nor very consistent. 2) That the single-parameter constructor of Lock blocks while the single-parameter constructor of TryLock doesn't is neither obvious nor very consistent. 3) Part of the reason for the above inconsistencies is that it's not clear whether the constructors of TryLock should block or not. 4) The TimedLock constructors have no way to request a non-blocking lock. To address these problems, I would like to redefine the lock concept constructors as follows (changes indicated with "^"): Definitions ----------- L: lock type l: lock variable (type L) m: mutex variable (appropriate Mutex type) s: initial lock state (enum type {NO_LOCK=0, LOCK=1}) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ b: block? (enum type blocking_state {NON_BLOCKING=0, BLOCKING=1}) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Lock Concept ------------ L l(m,s) //Blocking lock if s L l(m) //Blocking lock TryLock Concept --------------- L l(m,s,b) //Lock if s; blocking if b ^ ^^^^^^^^^^^^^ L l(m,b) //Lock; blocking if b ^ ^^^^^^^^^^^^^ TimedLock Concept ----------------- L l(m,s,b) //Lock if s; blocking if b ^ ^^^^^^^^^^^^^ L l(m,t) //Blocking lock, failing if not obtained by time t Note: s and b are defined as enums rather than bools to prevent existing code from breaking silently: e.g., if s and b were bools, "TryLock l(m, false)", which currently means "construct a lock object but don't obtain a lock", would change with the new definitions to mean "construct a lock and make a non-blocking attempt to obtain a lock". Also, enums are more explicit than bools. Note: these changes will cause some existing code to break by producing compile errors; the compile errors can be fixed by appropriately adding or changing parameters being passed to the changed TryLock and TimedLock constructors. Mike

2 1