October 2005 - Boost-users - lists.preview.boost.org

thread: condition::notify() does not require a lock - why?
by Pete Chapman 26 Oct '05

26 Oct '05

condition::wait requires that a lock is acquired prior to calling it, but condition::notify does not. The Java threading API (which in my limited experience seems similar to boost::thread) requires that a lock is acquired prior to calling Object.notify(). Is this potentially a problem? For example: scoped_lock myLock(myMutex); while (!some_variable) { // Another thread now sets some_variable = true, and // calls myCondition.notify() // then... // will this wait indefinitely? myCondition.wait(myLock); } Clearly if the other thread in this example were forced to obtain a lock on myMutex before calling myCondition.notify(), there would be no danger. In practice this is what I do anyway, but the boost:thread API doesn't force me to. Why not? Is there a potential problem if I don't? Many thanks...

2 2

bind with boost::function
by Eric Pruneau 26 Oct '05

26 Oct '05

Hi Here is my problem. I need to specify a callback function for use with OpenCV cvcam dll. I modified OpenCV to accept a boost::function1<void,IplImage*> instead of a simple void pointer to a function. The function is called by openCv on every frame of a video stream. Im trying to have the callback function member of a class e.g Class A { ... void init() { ... boost::function1<void,IplImage*> Myfunct; Myfunct = boost::bind(&A::Callback, this); cvcamSetcallback(Myfunct); // OpenCV function to set the callback } void Start() { /* tell opencv to start grabbing frames and to call Callback on every frame */ } void Callback(IplImage* img) { /* process the image */ } ... }; Its not like I want to directly call the callback function and use it like a functor. I want OpenCv to call my member function. When I try to compile this it give me: error C2091: function returns function on a C++ .net compiler. Im really confused and I dont have a clue about how to solve this. Thanks in advance Eric _________________________________________________________________ Balayez vos courriels entrants et sortants et les pièces jointes et contribuez à éliminer les virus destructeurs susceptibles dy être intégrés. http://join.msn.com/?pgmarket=fr-ca&page=features/virus Commencez dès maintenant à profiter de tous les avantages de MSN Premium et obtenez les deux premiers mois GRATUITS*.

2 1

[iostreams] Patch for exception message in mapped_file
by kml＠elreki.net 26 Oct '05

26 Oct '05

Hi! I noticed there's a bug in the detail::cleanup_and_throw() function. It doesn't use the error message passed in the arguments, but always reports "failed getting file size". Please find attached a patch against the CVS version. Cheers, Kim

2 1

[variant] type ambiguity help
by Aaron Griffin 25 Oct '05

25 Oct '05

This isn't a full blown issue really, but it's probably a lack of understanding on my part. variant< typeA, typeB, int, double> v; gives me conversion ambiguities. I can remove the int and it works fine. In an attempt to kludge it, I changed the int to a rational<int>. However, now, "v = 12" is converted to a double, not to the rational. First question, would be, how exactly is the variant choosing to use the double instead of the rational? Does it have anything to do with order? Would <double,rational<int> > be different compared to <rational<int>,double> ? Is there any way to "slay" this ambiguity. It's a rather moot point, but I'd like to know if it's an int or a double, seperately, not one single double which would require me to determine if it is a whole number or not. Thanks in advance, Aaron

1 1

date_time:
by __PPS__ 25 Oct '05

25 Oct '05

Are there better conversion functions from ISO date format (calendar, ordinal, week dates) in basic and/or extended format? Should conversion from invalid iso date like "20050231" fail or produce same results as date(2005,02,31). (I know that it produces the same results, but is it right to convert invalid iso date representation to valid date?) Another thing I noted, some conversion functions accept const std::string &, but most of them accept plain std::string, isn't better to convert all of them to constant references??? If anybody interested, I wrote such function that accepts iso date string in any of the above mentioned 6 combinations and returns date. It runs 90-100 times faster than equivalent date_from_iso_string from boost. And yes, it does make sure that "20050229" fails

4 6

Problems recognizing directionality of an edge
by Jeffrey Holle 24 Oct '05

24 Oct '05

I am using boost 1.33.0. I'm the guy who is working with spanning trees. I've decided that two graphs are in order. 1. The data graph, of type Graph, is bidirectional. 2. The tree graph, of type Tree, is undirectional. Tree is undirectional because this is the only way that I get expected results from a dfs. However, the direction of the associated edge needs to be recorded as a property in the Tree graph. I'm having difficulty in initializing this property. The problem is that when I create the spanning tree, all I have is an Graph::edge_descriptor. I find that an edge will always be in the target vertex's in_edges container, and in the source vertex's out_edges container. This don't allow identification of the edge direction. Am I missing something?

2 1

[boost-users] [MPL.Math] More major changes
by Cromwell Enage 24 Oct '05

24 Oct '05

A new directory, boost/mpl/math/unsigned_big_integral_, holds everything pertaining to unsigned_big_integral. The corresponding tag and forward declaration also have their own header files now. Also, enlarge_integral and its forward declaration have their own files as well. This should maximize decoupling of the three entities. The big_integral.cpp example and test programs have been updated to reflect these changes. As for double, whole_part and rational_part have been improved, but you'll need to #include unsigned_big_integral and rational to use them. Major limitation: attempting to calculate the rational_part of any double smaller than 1e+30 will most likely break your compiler's back; unsigned_big_integral and rational still won't mix. Play around with the double.cpp example program at your own risk. Cromwell D. Enage __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com

1 0

[date_time] problem adding negative number of months
by Guido Spahn 24 Oct '05

24 Oct '05

Hello, when adding a negative number of months to a date(y, m, d), an exception "Month number is out of range 1..12" is raised if the resulting month m2 belongs to a year y2 < y and m2 > m. For example date(2005, Jan, 1) + months(-1) fails date(2005, Feb, 1) + months(-1) doesn't fail date(2005, Feb, 1) + months(-2) fails date(2005, Feb, 1) + months(-12) doesn't fail date(2005, Feb, 1) + months(-13) doesn't fail date(2005, Feb, 1) + months(-14) fails I checked it only with release 1.33 using the appended code. There is no problem adding or substracting any positive number of months. Hopefully I didn't miss this topic in the mailing list archive. Thanks in advance, Guido #include <boost/date_time/gregorian/gregorian.hpp> int main() { using namespace boost::gregorian; date d(2005, May, 1); for (long m = 0; m >= -40; --m) { try { std::cerr << d << " + (" << m << "): "; std::cerr << d + months(m) << std::endl; } catch (std::exception& e) { std::cerr << "FAILED: " << e.what() << std::endl; } } return 0; }

3 3

compiler serialization library failed using -sTOOLS=msvc option
by Lee FireStorm 23 Oct '05

23 Oct '05

I have do it in linux as3.0, and all is ok. While I had to compiler boost 1.33 under windows and use the older compilers - MSVC 6. All env set, I think , is well. That is: SET MSVC_ROOT="c:\Program Files\Microsoft Visual Studio\VC98" SET VISUALC="c:\Program Files\Microsoft Visual Studio\VC98" PATH=d:\boost_1_33_0;d:\boost_1_33_0\tools\build\jam_src\bin.ntx86;%PATH% set SPIRIT_ROOT=c:/spirit16 I successfully got some libs as data_time, thread ,regex, etc. Except serialization. I download spirit-1.6.1,spirit-only-1.6.1,spirit-1.6.3,spirit-only-1.6.3. I tried everyone of these four, and all failed. But I really need the libs of serialization such as libboost_serialization-vc6-sgd-1_33.lib, Anyone knows how to compile these libs?

2 1

[serialization] serializing an stl container.
by Lucas Meijer 23 Oct '05

23 Oct '05

Hey, I've been banging my head against the wall for hours on this one. I'd like to serialize stl containers to an archive. Starting simple: a vector<int> This is what I think I should do: 1) include the proper support header file for the to be serialized container. #include <boost/serialization/vector.hpp> 2) create an archive, can be any type: std::stringstream ss; boost::archive::text_oarchive oa(ss); 3) operator<< my container into this archive: const vector<int> myvec; oa << myvec; The "oa << myvec" generates lots of warnings, and when running it anyway, it crashes in smart_cast.hpp I've tried lots of containertypes, and creating them in other ways than shown above, using a binary_oarchive, but it doesn't seem to make a difference. (int myint; oa << myint; works fine) Does anybody see what I'm missing here? Thanks, Lucas

2 9