- Boost-users - lists.preview.boost.org

Error with boost::regex library in Borland C++ Builder 6
by Giany2k 14 Mar '02

14 Mar '02

Hi, Can anyone explain to me why I get "Abnormal program termination" at the exit of the function shown below in the example. The error occurs when exiting the function after it was called second time. //---------------------------------------------------------------------- ----- #include <vcl.h> #pragma hdrstop #include <boost/regex.hpp> #include <string> #include "Unit1.h" //---------------------------------------------------------------------- ----- #pragma package(smart_init) #pragma resource "*.dfm" TForm1 *Form1; //---------------------------------------------------------------------- ----- __fastcall TForm1::TForm1(TComponent* Owner) : TForm(Owner) { Edit1->Clear(); } //---------------------------------------------------------------------- ----- using namespace boost; using namespace std; void __fastcall TForm1::Button1Click(TObject *Sender) { Memo1->Clear(); if ( Edit1->Text.IsEmpty() ) return; regex e( std::string(Edit1->Text.c_str()) ); cmatch what; string response; TFileStream* file = new TFileStream( "1.htm", fmOpenRead ); response.resize(file->Size+1); file->Read(response.begin(), file->Size); response[file->Size] = 0; delete file; std::string::const_iterator start; start = response.begin(); while( regex_search( start, what, e ) ) { for ( unsigned int i = 0 ; i < what.size() ; i++ ) { std::string result = std::string(what[i].first, what[i].second); Memo1->Lines->Add( result.c_str() ); } Memo1->Lines->Add(""); start = what[0].second; } } //---------------------------------------------------------------------- -----

1 0

Build problems on Linux with gcc 3.0.4
by Peter Simons 13 Mar '02

13 Mar '02

Hi, I have tried to build the current boost library as checked-out of the CVS repository, but I failed miserably. For one, the build process failed to recogize that Python is installed in /usr/local/python rather than /usr/local. As a result, it didn't find the python-specific include files either. I tried to set the appropriate environment variables but failed. So I added the directory /usr/local/python/include/python2.2 to CPLUS_INCLUDE_PATH and then it worked -- at least a little better than before. Once the Python includes were found, the build process progressed to actually linking some programs: gcc-Link-action status/bin/config_test.test/gcc/debug/runtime-link-dynamic/threading-multi/config_test status/bin/config_test.test/gcc/debug/runtime-link-dynamic/threading-multi/config_test.o: In function `boost_has_clock_gettime::f()': /home/simons/projects/boost-current/libs/config/test/boost_has_clock_gettime.cxx:19: undefined reference to `clock_gettime' collect2: ld returned 1 exit status Another failure was: ...failed gcc-Link-action status/bin/config_test.test/gcc/release/runtime-link-dynamic/threading-multi/config_test ... gcc-C++-action status/bin/class_concept_fail_expected.test/gcc/debug/runtime-link-dynamic/class_concept_fail_expected.o /home/simons/projects/boost-current/boost/concept_check.hpp: In member function `void boost::EqualityComparableConcept<TT>::constraints() [with TT = foo]': /home/simons/projects/boost-current/libs/concept_check/class_concept_fail_expected.cpp:25: instantiated from here /home/simons/projects/boost-current/boost/concept_check.hpp:265: no match for `foo& == foo&' operator /home/simons/projects/boost-current/boost/concept_check.hpp:266: no match for `foo& != foo&' operator Plenty of files failed to compile or to link, but I don't want to flood the list with error messages. So, first of all, is there any "right" way to tell the build process where python is installed? Maybe I should start from the scratch from there. -peter

1 0

Jam: Building libraries that use the static runtime library
by Dale Peakall 13 Mar '02

13 Mar '02

I need to build libraries that link against the static runtime library: I tried the following Jam invocation: jam -sBOOST_ROOT=. -sTOOLS="msvc-stlport" -sBUILD="debug release <runtime-link>static" But jam balks and reports: Error: <libs!python!build>libboost_python_static: target requirements conflict for requested build { <runtime-link>dynamic <target-type>LIB <runtime-link>static } Help! So, I've obviously got it wrong (btw I'm not interested in building python library just threads and regex). TIA. - Dale.

2 1

AIX and Boost 1.27.0 (Does not compile)
by samsrd 12 Mar '02

12 Mar '02

Has anyone tried to compile Boost 1.27.0 (or any version for that matter) on IBM AIX using the xlC compiler (IBM Visual Age). There are many compile errors (usually template related) along the lines of: "../../../../ThirdParty/boost/regex/detail/regex_cstring.hpp", line 77.40: 1540- 0152 (W) A template dependent name that is a type must be qualified with "typena me". "../../../../ThirdParty/boost/regex.hpp", line 528.55: 1540-0152 (W) A template dependent name that is a type must be qualified with "typename". "../../../../ThirdParty/boost/regex.hpp", line 532.13: 1540-1628 (S) Every templ ate parameter for a constructor template must be used in the parameter list of t he constructor. "../../../../ThirdParty/boost/regex.hpp", line 532.34: 1540-0152 (W) A template dependent name that is a type must be qualified with "typename". "../../../../ThirdParty/boost/regex.hpp", line 744.13: 1540-0152 (W) A template dependent name that is a type must be qualified with "typename". "../../../../ThirdParty/boost/regex.hpp", line 938.4: 1540-0152 (W) A template d ependent name that is a type must be qualified with "typename". "../../../../ThirdParty/boost/regex.hpp", line 438.27: 1540-0130 (S) "def_alloc_ param_traits" is not declared. "../../../../ThirdParty/boost/regex.hpp", line 1414.1: 1540-0062 (S) The incompl ete class "match_results" must not be used as a qualifier. "../../../../ThirdParty/boost/regex.hpp", line 1455.35: 1540-0140 (S) The text " &" is unexpected. "match_results" may be undeclared, ambiguous or may require " typename" qualification. "../../../../ThirdParty/boost/regex.hpp", line 444.7: 1540-1303 (I) "template cl ass match_results" is not visible. @

2 1

thread::yield() vs Sleep()ing
by simonwlb 10 Mar '02

10 Mar '02

I'm using thread::yield() in my multi-threaded cross-platform application to avoid "CPU burn". This is because I have threads that are in an infinte loop in order to do some polling. There are no mutexes to wait on in the loop. void X::threadFunction() { for(;;) { if( m_x->isReady() ) doSomething(); . . . boost::thread::yield(); } } If the answer is "you should almost never poll like that - try to use mutexes etc. as much as possible" please say so (I'm a newbie to threads)! In any case, if I choose to have a thread in an infinite loop anyway, I have three options (assuming I have no mutexes available to me): 1. Don't yield or Sleep at all. 2. Sleep for a while - say for example, 10 ms. 3. Yield. Option 1 is clearly wrong. The CPU usage will go to 100% and the whole machine may slow down to an unusable speed. Option 2 works well in my emperical tests. The thread loop executes plenty fast enough and my Windows CPU meter shows virtually 0% CPU usage. 10ms is a bit of an arbitrary figure though - My code needs to work on a variety of software and hardware platforms, so the ideal number is hard or impossible to know. Option 3 also works well in my emperical tests and seems like the neat solution (no arbitrary waiting period). My application and the system as a whole both run smoothly and fast. The only irksome thing is that the CPU meter shows 100% usage. That's a bit alarming for me and the users of my software. My questions: 1. Should thread loops without logical wait conditions be avoided as much as possible? Is that the problem? 2. Should I just yield() or should I pause for a time period? 3. How do you select a time period to wait? Simon Bailey

4 4

Boost Building
by karl_m_au 09 Mar '02

09 Mar '02

Hiya I'm trying to build boost 1.27.0 on a Debian linux system. I've built the boost version of jam. Both g++ 2.95.3 and g++ 3.0.3 are installed. I'm pretty sure I've made the appropriate changes to get g++ 3.0 to be used in the build process. I've used various versions of the command line recommended in the documentation. Very few of the builds work, becuse of this: g++-3.0 -c -Wall -ftemplate-depth-100 -I"libs/regex/build" -o "libs/regex/build/bin/libboost_regex.a/gcc-3/debug/c_regex_traits.o" "libs/regex/build/../src/c_regex_traits.cpp" Somehow the <include> directive isn't working. Any clues? TIA Karl M

3 4

Boost python
by andrewperella 08 Mar '02

08 Mar '02

I am having great trouble with boost python whenever I do anything slightly more compilcated than the examples. I have a wrapped a Vector3 class with 3 floats x,y,z. In python: a=m2.Vector3(1,2,3) a.x 1.0 a.x=100 a.x 100.0 This is correct - ie I can disply and modify a.x Unfortunatedly if wrap another class - Camera which contains a member variable : Vector3 pos then I cannot set camera.pos.x Looking in the debugger, I seem to be getting a copy (in fact sometime 3 or more copy constructors are called) of pos and I am modifying this temporary copy! Help. Surely I can wrap no pointer member variables? Regards, Andrew

1 0

1.27 BGL and VC++7.0 compilation problem
by elgatomoses 07 Mar '02

07 Mar '02

I am having trouble compiling the Graph library with VC++7.0 (.NET) - I'm getting "INTERNAL_COMPILER_ERROR". The problem has to do with the is_array definition in composite_traits.hpp. Has anyone else encountered this problem - perhaps there are some compiler switches that I can set... TIA Mark Kruger

2 1

Boost.build question: Adding a new unit-test rule
by Helge Penne 04 Mar '02

04 Mar '02

Hi. I'm working on a project where some of the unit test are very inconvenient to implement with CppUnit and Mock classes or stubs alone, but would be very simple to implement by having them print messages to standard output, pipe this to a log file and diff this against a reference file. I added the following to our Jamrules file (most of it is stolen from the standard unit-test rule and run action): rule unit-test-diff { local files = [ exe $(<) : $(2) : $(3) : $(4) : EXE ] ; type-DEPENDS test : $(files) ; local file ; for file in $(files) { run-and-diff $(file) ; } } actions run-and-diff { $(SHELL_SET)PATH=$(RUN_PATH) $(SHELL_EXPORT)PATH $(<) > $(<:D=:S=.log) diff $(<:S=.log) $(<:S=.ref) # diff $(<:D=:S=.log) $(<:D=:S=.ref) } This runs the unit test program, pipes the output to <targetname>.log and diffs agains <targetname>.ref. There is only one problem: I cannot get Boost.build to locate these files in the correct folder. The Jamfile for the unit test is located in the same folder as the unit test source code. This is the only natural place for the ref-file as well. The action above has three different alternatives for the actual diff operation. The first uses the full path of the executable, but replaces the last part of the file name with .log and .ref. This places them in the binary folder of the unit test program, which isn't not very convenient. The second strips away the path, but since the currect folder always is the project root (not the folder where the Jamfile is), this places the log and ref files on the project root, which is almost as bad as the first alternative. I also experimented briefly with :G but this has no effect as the file path did not seem to contain grist. One possible solution might be to add a path feature to the target and set this to the reference file name, but I'm not quite sure how to implement this for rules that are not part of a tools.jam-file I really need the unit test rule to place these files in the folder where the rest of the test source code is. Does anyone know how to do this? - Helge Penne

3 2

(no subject)
by Hook.RobertA(CC) 04 Mar '02

04 Mar '02

I put the following up at source forge support, then realised that doesn't seem to be likely to get a response.... I would have put this through the yahoo group, but I'm stuck behind a mind-bogglingly repressive firewall. SO I'll scream for help here, bearing in mind the dictum that in cyberspace no-one can hear you scream. I have download 1.27.0, and am attempting to build and use Boost under HP-UX 11.00. There are two problems, and I cannot tell if they are related or not. Firstly, the install instructions along the lines of jam -sBOOST_ROOT=. -=sTOOLS="hpux" simply produce the following (after I bodgied up a hpux-tools.jam) +: line 1304: yacc stack overflow at keyword } don't know how to make all ...found 1 target... ...can't find 1 target... after smacking my head on the desk for several days, I've given up on that one and attempted to build boost_1_27_0/libs/tuple/test/io_test.cpp using aCC -c +z +p $APP_INCLUDE -o io_test.o io_test.cpp (where APP_INCLUDE is -I <yada/yada/yada> as appropriate) which results in "tuple_basic.hpp", line 290 # Cannot generate a specialization of incomplete template struct STATIC_ASSERTION_FAILURE<(bool)0>. BOOST_STATIC_ASSERT(length<cons>::value == 2); // check length = and errors cascading after that. Looking at various messages and comments, I believe that someone somewhere has managed to get boost working under HP-UX, but I'm dammned if I can. Help! Oh yes, by the way, I'm stuck behind a repressive firewall and cannot even *read* the message archive at Yahoo, let alone subscribe / post! (Puts head down on desk and begins sobbing quietly) -- Robert Hook - DCC Analyst/Programmer Polaris Group, Information Systems Branch Queensland Police Service Hook.RobertA(a)police.qld.gov.au PH: 07 3364 3807 ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify securityscanner(a)police.qld.gov.au. This footnote also confirms that this email message has been checked for the presence of computer viruses. **********************************************************************

1 0

[Fwd: compile time container type identification]
by hicks 01 Mar '02

01 Mar '02

Hello I need to identify if a container is of a certain type. Specifically, from inside the code for container template <class T> Cont, I want to identity if T is itself of type Cont<U> for some U. (The type of U does not matter.) I couldn't see anything in *_traits.hpp that looks applicable. Any advice or references would be appreciated. Craig Hicks KGK Tokyo [Non-text portions of this message have been removed]

1 0

RE: [Boost-Users] Linking an Archive
by Tom Matelich 27 Feb '02

27 Feb '02

The first getting_started1.so contains all the symbols from all of the objects linked into it. The second one only contains the symbols from getting_started.o and the symbols from libboost_python.a which satisfy unsatified symbols of getting_started1.o. Linking shared library is like linking an executable, it only takes the symbols from archives that it needs to satisfy itself. The first question is why you want libboost_python.a. If it is a valid reason, then you need to find a way to get all the code you need out of it into the shared lib. The first would be to use ar to get all the objects out, then take those and put them into the shared library. The second would be to keep adding unsatified symbols to getting_started.o until getting_started1.so has everything it needs. Tom > -----Original Message----- > From: doak87508 [mailto:jdoak@lanl.gov] > Sent: Wednesday, February 27, 2002 10:21 AM > To: Boost-Users(a)yahoogroups.com > Subject: [Boost-Users] Linking an Archive > > > I am playing around with using an archive to created a shared library > (.so file) instead of linking all the individual objects. I have > success when I link individual objects: > > % g++ -shared classes.o conversions.o errors.o extension_class.o > functions.o init_function.o module_builder.o objects.o types.o > cross_module.o getting_started1.o -o getting_started1.so > > % python > 1;31mh[1] >>> import getting_started1 > 1;31mh[1] >>> getting_started1.greet() > 'hello, world' > > Now, I create an archive: > > % ar r libboost_python.a classes.o conversions.o errors.o > extension_class.o functions.o init_function.o module_builder.o > objects.o types.o cross_module.o > > Then, I link that archive into the shared library: > > % g++ -shared libboost_python.a getting_started1.o -o > getting_started1.so > > % python > 1;31mh[1] >>> import getting_started1.so > Traceback (most recent call last): > File "<stdin>", line 1, in ? > ImportError: ./getting_started1.so: undefined symbol: > add__Q35boost6python19module_builder_basePQ45boost > 6python6detail8functionPCc > > I have also tried linking as follows with no change in result: > > % g++ -shared -L. -lboost_python getting_started1.o -o > getting_started1.so > > Does someone know how to get linking working with an archive file? > > JD > > > > Info: <http://www.boost.org> > Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> > Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> > > > Your use of Yahoo! Groups is subject to > http://docs.yahoo.com/info/terms/ > >

2 1

in-place set_union/difference/etc ...
by ivan_dobrianov 27 Feb '02

27 Feb '02

Does anyone know of any existing standard tool [an insert iteartor or something equivalent] that can allow one to perform union, difference, etc on sets and maps in-place. E.g. instead of using set<MyClass> s1, s2, temp; set_union( s1.begin(), s1.end(), s2.begin(), s2.end(), inserter(temp, temp.begin())); temp.swap(s2); i'd like to do, say: copy(s1.begin(), s1.end(), my_unioner(s2, s2.begin())); which will do s2 = s1 + s2 in LINEAR time. It's pretty simple to write an inserter like that, but I am just wondering if there is one already ... or something similar. Thanks!

1 0

Bug in boost_1_27_0.tar.gz threads library
by Martin Brown 27 Feb '02

27 Feb '02

Hi There seems to be a missing '}' in libs/thread/src/once.cpp when compiling on gcc-3.04 on Redhat Linux 6.2. I added one and it seems to work. Not sure if it is in the right place. Context diff: *** once.cpp Thu Feb 7 15:33:15 2002 --- once.cpp.new Wed Feb 27 15:53:02 2002 *************** *** 55,62 **** --- 55,63 ---- { once_callback* cb = reinterpret_cast<once_callback*>(pthread_getspecific(key)); (**cb)(); } + } #elif defined(BOOST_HAS_MPTASKS) void *remote_call_proxy(void *pData) { std::pair<void (*)(), boost::once_flag *> &rData(*reinterpret_cast<std::pair<void (*)(), boost::once_flag *> *>(pData)); Thanks for some great libraries! Martin __________________________________________________ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com

2 1

Linking an Archive
by doak87508 27 Feb '02

27 Feb '02

I am playing around with using an archive to created a shared library (.so file) instead of linking all the individual objects. I have success when I link individual objects: % g++ -shared classes.o conversions.o errors.o extension_class.o functions.o init_function.o module_builder.o objects.o types.o cross_module.o getting_started1.o -o getting_started1.so % python 1;31mh[1] >>> import getting_started1 1;31mh[1] >>> getting_started1.greet() 'hello, world' Now, I create an archive: % ar r libboost_python.a classes.o conversions.o errors.o extension_class.o functions.o init_function.o module_builder.o objects.o types.o cross_module.o Then, I link that archive into the shared library: % g++ -shared libboost_python.a getting_started1.o -o getting_started1.so % python 1;31mh[1] >>> import getting_started1.so Traceback (most recent call last): File "<stdin>", line 1, in ? ImportError: ./getting_started1.so: undefined symbol: add__Q35boost6python19module_builder_basePQ45boost 6python6detail8functionPCc I have also tried linking as follows with no change in result: % g++ -shared -L. -lboost_python getting_started1.o -o getting_started1.so Does someone know how to get linking working with an archive file? JD

1 0

autoconf+make
by josefsyk 26 Feb '02

26 Feb '02

I've been struggling to port parts of the boost libraries to Linux using gnu's automake, just to hit the wall every time. Anyone out there who has done this already? The files needed are: boost/smart_ptr.hpp boost/pending/ct_if.hpp boost/random.hpp boost/bind.hpp boost/function.hpp Would be nice to have the Makefile.am for BGL too. cheers.

2 1

boost::condition question
by svanechel 26 Feb '02

26 Feb '02

Hi, I'm trying to create a mechanism that synchronizes access to shared data on a per-operation basis. The idea is that some operations (E.g. read operations) can proceed concurrently, while others (write operations) must block all other threads. I thought boost::condition would do the trick. Since I do not have that much experience in multithreaded programming it is entirely possible that I do not fully understand boost::condition or its allowable usage. I tried to put operations in an "Operation Group". All operations belonging to the same group are allowed to execute concurrently, while all others are blocked (E.g. all read operations are in the same group). It is also possible that an operation doesn't specify an Operation Group, in which case a temporary group is assigned to the operation for each thread that calls it. Write operations are constructed this way. Below is an example of what I mean (OperationLevelLockable is defined below): class SomeClass : private OperationLevelLockable { public: int ReadData() { Lock lock( this, READ_OP_GROUP ); return ( mData ); } void WriteData( int data ) { Lock lock( this ); // assign temp group to operation, // so each write // operation blocks all other accesses mData = data; } private: int mData; }; I'm developing on Windows 2000 on a multiprocessor machine. This is the code I use to test my idea: const int32 OPERATION_GROUP_1 = 1; const int32 OPERATION_GROUP_2 = 2; const int NUM_THREADS = 4; class TestOperationLevelLockable : public MTTestCase { public: TestOperationLevelLockable() : mShared( this ), mCurrentGroup( 0 ), mThreadCount( 0 ) { // Create two sets of threads, each performing a // different operation // When the TestOperationLevelLockable object is // constructed all // registered threads are created an start running RegisterThread( this, &TestOperationLevelLockable::Operation1Thread, NUM_THREADS ); RegisterThread( this, &TestOperationLevelLockable::Operation2Thread, NUM_THREADS ); } void CallFromGroup( int32 opGroup ) { boost::mutex::scoped_lock lock( mMutex ); if ( 0 == mCurrentGroup ) { mCurrentGroup = opGroup; } if ( ( mCurrentGroup != opGroup ) && ( mThreadCount < NUM_THREADS ) ) { ReportFailure( "OperationLevelLockable failed to lock out operations of different group", __FILE__, __LINE__ ); } ++mThreadCount; } protected: class MTObject : public OperationLevelLockable { public: MTObject( TestOperationLevelLockable* target ) : mTarget( target ) {} void CallFromGroup( int32 opGroup ) { mTarget->CallFromGroup( opGroup ); } private: TestOperationLevelLockable* mTarget; }; MTObject mShared; int32 mCurrentGroup; int mThreadCount; boost::mutex mMutex; void DoMTTest() { } void Operation1Thread() { MTObject::Lock lock( mShared, OPERATION_GROUP_1 ); mShared.CallFromGroup( OPERATION_GROUP_1 ); SleepThread( 30 ); // seconds, to keep the thread busy } void Operation2Thread() { MTObject::Lock lock( mShared, OPERATION_GROUP_2 ); mShared.CallFromGroup( OPERATION_GROUP_2 ); SleepThread( 30 ); // seconds, to keep the thread busy } }; And here is OperationLevelLockable: class OperationLevelLockable { class AccessCounter; public: OperationLevelLockable(){} typedef Internal::DefaultTMIntType IntType; template<typename T> struct MakeSharedType { typedef T volatile Type; }; class Lock; friend class Lock; class Lock { public: Lock( OperationLevelLockable& obj ) : mCounter( obj.mOpAccessCounter ) { mCounter.EnterOperation( Lock::TempOpGroup() ); } Lock( OperationLevelLockable& obj, int32 opGroup ) : mCounter( obj.mOpAccessCounter ) { mCounter.EnterOperation( opGroup ); } ~Lock() { mCounter.LeaveOperation(); } private: Lock( Lock const& ); Lock& operator=( Lock const& ); static IntType TempOpGroup() { static volatile IntType opGroup = Internal::kFIRST_AUTO_GROUP; IntType newGroup = OperationLevelLockable::AtomicIncrement( opGroup ); if ( newGroup >= Internal::kLAST_AUTO_GROUP ) { OperationLevelLockable::AtomicAssign( opGroup, Internal::kFIRST_AUTO_GROUP ); } return ( newGroup ); } AccessCounter& mCounter; }; static inline IntType AtomicIncrement( volatile IntType& lval ); static inline IntType AtomicDecrement( volatile IntType& lval ); static inline void AtomicAssign( IntType& lval, IntType rVal ); static inline void AtomicAssign( volatile IntType& lval, IntType rVal ); private: class AccessCounter : private boost::base_from_member<boost::recursive_mutex> { public: AccessCounter() : mLock( member ), mRefCnt( 0 ), mOpGroup( Internal::kANY_GROUP ) {} bool IsOpGroupActive( IntType opGroup ) const { return ( ( Internal::kANY_GROUP == mOpGroup ) || ( opGroup == mOpGroup ) ); } void LeaveOperation() { if ( 0 == OperationLevelLockable::AtomicDecrement( mRefCnt ) ) { SetOpGroup( Internal::kANY_GROUP ); mCanOpProceed.notify_all(); } } void EnterOperation( IntType opGroup ) { mCanOpProceed.wait( mLock, CanOpProceed( *this, opGroup ) ); SetOpGroup( opGroup ); OperationLevelLockable::AtomicIncrement( mRefCnt ); } private: AccessCounter( AccessCounter const& ); AccessCounter& operator=( AccessCounter const& ); struct CanOpProceed { CanOpProceed( AccessCounter const& counter, IntType opGroup ) : mOpGroup( opGroup ), mCounter( counter ) {} bool operator()() const { return ( mCounter.IsOpGroupActive( mOpGroup ) ); } private: CanOpProceed& operator=( CanOpProceed const& ); int32 mOpGroup; AccessCounter const& mCounter; }; void SetOpGroup( IntType opGroup ) { OperationLevelLockable::AtomicAssign( mOpGroup, opGroup ); } boost::recursive_mutex::scoped_lock mLock; boost::condition mCanOpProceed; volatile IntType mRefCnt; volatile IntType mOpGroup; }; OperationLevelLockable( OperationLevelLockable const& ); OperationLevelLockable& operator=( OperationLevelLockable const& ); AccessCounter mOpAccessCounter; }; The idea is that the first operation that is executed acquires the lock and blocks out all threads not belonging to the same Operation Group. The original blocking works fine. Only threads executing in the same Operation Group are allowed to continue. When the last thread of the group leaves the operation in OperationLevelLockable::AccessCounter::LeaveOperation it releases all waiting threads and a thread from the other group is allowed to run. However when this thread finishes no new threads are allowed to run anymore and the application hangs. Am I using boost::condition incorrectly. Should I use something else? It the test code wrong? Are there any other errors in the code that prevent it from behaving as expected. Should I use another technique to achieve the same? I hope there is someone out there who has the hart to read through the entire posting and help me. I hope I have given enough information. Thx. Sven Van Echelpoel.

2 4

Building regex with msvc6sp5 and stlport
by John Maddock 25 Feb '02

25 Feb '02

>I have succesfully built the regex library >under msvc6sp5, using the vc6-sltport.mak file. > >But I notice that the compiler gives me warnings >relating to the microsoft stl include directories: > >example: vc98\include\iterator: blah blah blah Then you aren't compiling / linking to STLport >Should it not be linking to the stlport include files? Yes. Did you either * update your INCLUDE and LIB environment variables to point to STLport. or * invoke the makefile using something like: nmake CXXFLAGS="-Ic:\STLport-4.5.1\stlport" -fvc6-stlport.mak - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

2 1

Thread support?
by John Maddock 24 Feb '02

24 Feb '02

>If I'm using Boost in a single threaded app, do I need to define >BOOST_DISABLE_THREADS to avoid Boost using locks, etc. internally? You don't need to, but it may remove some redundent code, if your app is single threaded *and* links to a multithreaded runtime. - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

4 4

Problem linking with regex on Win32
by John Maddock 23 Feb '02

23 Feb '02

>Run-time Library: Multithreaded DLL/Debug Multithreaded DLL > >I've got a couple of problems trying to get my app to link against regex. > >I've tried linking to the debug and release versions of both the DLL and >non-DLL versions of the regex library and in each case I get errors about >unresolved external symbols for w32_regex_traits. > >w32_regex_traits.obj has been produced by Jam. > >In addition the static version winges about lots of locally defined symbols. > >I've specified BOOST_REGEX_NO_LIB as the generated filenames are not what >Jam produces. There could be a number of problems: Where the libs compiled with STLport (even a small error in the STLport include path could have messed this up)? The debug STLport libs are probably compiled by jam with __STL_DEBUG defined, your source will need to set this also. If you want to check your jam compiled libs, get a dump of the symbols they contain - if they're using the Dinkumware std lib then you'll see them referencing std::basic_string, release STLport will reference _STL::basic_string, and debug STLport will reference _STLD::basic_string. Failing that just create a static lib project in your IDE and add the regex sources to that with the build setting you want (this is probably the quickest solution IMO). Finally define BOOST_REGEX_STATIC_LINK to suppress the import/export warnings - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

1 0

Building regex with msvc6sp5 and stlport
by boxerab 22 Feb '02

22 Feb '02

Hello, I have succesfully built the regex library under msvc6sp5, using the vc6-sltport.mak file. But I notice that the compiler gives me warnings relating to the microsoft stl include directories: example: vc98\include\iterator: blah blah blah Should it not be linking to the stlport include files? I have added the stlport directory location into the mak file, and I have it added to the list of include directories via the compiler gui. Thanks. Aaron Boxer boxerab(a)NOSPAMyahoo.com

1 0

Iterators
by Colin Fox 21 Feb '02

21 Feb '02

This isn't a boost specific message per se, I guess, but I thought someone here may know the answer. I have a list<> of smart_ptrs. Some of the elements in the list are ready for processing under various circumstances. What I'd like would be a custom iterator that when I say "++i" it checks the element to see if it's ready, and if not, advance to the next, and so on until it finds a ready element. If it doesn't find one, then i == list.end(). This would simulate having a list that contained only ready elements, without having to have multiple lists. I have a number of books on C++, but I can't figure out from them how to create a custom iterator. Can anyone help? -- Colin Fox CF Consulting Inc. cfox(a)crystalcherry.com

2 1

[BGL] BGL Book Erratum
by Eric Eide 21 Feb '02

21 Feb '02

Page 12, Figure 1.2: the edge (a,d) should be black, and (b,d) should be gray. ...unless `breadth_first_search' is using the `delay_map' in some way that is not obvious to me. Eric. -- ------------------------------------------------------------------------------- Eric Eide <eeide(a)cs.utah.edu> . University of Utah School of Computing http://www.cs.utah.edu/~eeide/ . +1 (801) 585-5512 voice, +1 (801) 581-5843 FAX

2 1

const smart pointers
by nckgl 21 Feb '02

21 Feb '02

how do I define const smart pointer? for example for "int" type 1. scoped_ptr<const int> 2. const scoped_ptr<int> If the latter then it may be better to re-define get() as follows T* get(){ return px; } // never throws const T* get()const { return px; } // never throws Now scoped_ptr::get() is defined as follows T* get() const { return px; } // never throws The same apply to other accessors (operator-> and so on)

2 1

Problem linking with regex on Win32
by Dale Peakall 21 Feb '02

21 Feb '02

I'll try and add some content before sending a message this time ;) Platform: 2K Compiler: VC6 SP5 Library: stlport-4.0 Boost: 1.27.0 Build Tool for Boost: Jam Run-time Library: Multithreaded DLL/Debug Multithreaded DLL I've got a couple of problems trying to get my app to link against regex. I've tried linking to the debug and release versions of both the DLL and non-DLL versions of the regex library and in each case I get errors about unresolved external symbols for w32_regex_traits. w32_regex_traits.obj has been produced by Jam. In addition the static version winges about lots of locally defined symbols. I've specified BOOST_REGEX_NO_LIB as the generated filenames are not what Jam produces. Here's the full details: --------------------Configuration: request_line_test - Win32 Debug-------------------- Linking... http_request_line.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: static void __cdecl boost::w32_regex_traits<char>::transform_primary(class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<c har> > &,class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<char> > const &)" (__imp_?transform_primary@?$w32_regex_traits@D@boost@@SAXAAV?$basic_string@D U?$char_traits@D@std@@V?$allocator@D@_STL@@@_STL@@ABV34@@Z) http_request_line.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: static bool __cdecl boost::w32_regex_traits<char>::lookup_collatename(class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator< char> > &,char const *,char const *)" (__imp_?lookup_collatename@?$w32_regex_traits@D@boost@@SA_NAAV?$basic_string @DU?$char_traits@D@std@@V?$allocator@D@_STL@@@_STL@@PBD1@Z) http_request_line.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: __thiscall boost::bad_expression::bad_expression(class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<char> > const &)" (__ imp_??0bad_expression@boost@@QAE@ABV?$basic_string@DU?$char_traits@D@std@@V? $allocator@D@_STL@@@_STL@@@Z) http_request_line.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: static class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<char> > __cdecl boost::re_detail::w32_traits_base::error_strin g(unsigned int)" (__imp_?error_string@w32_traits_base@re_detail@boost@@SA?AV?$basic_string@DU ?$char_traits@D@std@@V?$allocator@D@_STL@@@_STL@@I@Z) http_request_line.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: static void __cdecl boost::w32_regex_traits<char>::transform(class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<char> > & ,class _STL::basic_string<char,struct std::char_traits<char>,class _STL::allocator<char> > const &)" (__imp_?transform@?$w32_regex_traits@D@boost@@SAXAAV?$basic_string@DU?$char_ traits@D@std@@V?$allocator@D@_STL@@@_STL@@ABV34@@Z) Debug/request_line_test.exe : fatal error LNK1120: 5 unresolved externals Error executing link.exe. request_line_test.exe - 6 error(s), 0 warning(s) Thanks, - Dale.

1 0