June 2011 - Boost-users - lists.preview.boost.org

Boost Atomic ?
by John M. Dlugosz 30 Jun '11

30 Jun '11

I was surprised to see that the Boost distribution doesn't have the "atomic" operations library. I expected it to be part of the Threading, modeled on the C++0x draft. I found <http://www.chaoticmind.net/~hcb/projects/boost.atomic/> via Google, from 2009. But N3225=10-0215 (is there a later Draft available ?) §29.2 shows lots of change bars, and the <atomic> header seems completely redone, it appears to make everything nicely templatated. I wonder if this is at all up-to-date, or why it's not already more officially part of Boost. Does anybody know? More to the point, I want to write some code that is portable across platforms, including emerging platforms like tablets which I expect will be more varied in their CPU architecture. I've been spoiled with writing x86-only for so many years! It makes sense to use the constructs that will be shipped with the compiler some day. So, where can I find that? —John

5 7

Question about fusion::map
by Frank Astier 29 Jun '11

29 Jun '11

Hi - I'm using fusion::map, and I've noticed that if I have: struct attr0 {}; typedef fusion::pair<attr0, double> Attr0; struct attr1 {}; typedef fusion::pair<attr1, double> Attr1; And then: typedef fusion::map<Attr0> ProfileType1; typedef fusion::map<Attr0, Attr1> ProfileType2; When I create N instances of ProfileType2, the process size at runtime is about twice the size of the same process creating N instances of ProfileType1. No big surprise, I guess, but if I *never* mention the type Attr1 in my code, is there a way to "remove" it at compile time from ProfileType2? I.e. I would want the compilation to remove the parts of the fusion::map that are guaranteed not to be used at runtime, so that I don't spend memory that I will never use. Is that possible? Thanks a lot! Frank fastier(a)yahoo.com

1 0

Re: [Boost-users] Using Boost::Bind
by David McCallum 29 Jun '11

29 Jun '11

I should clarify, this is what I want to achieve: boost::function<void (DBuffer&)> singleHandler; void addHandler(boost::function<void (DBuffer&)> newHandler) { singleHandler = newHandler; } The problem is that I don't know how to call boost::bind with a DBuffer parameter, where DBuffer is one of my own classes. > You will want to learn a bit more about Boost.Bind. You can bind your > pointer or variable at the bind call. Parameters at the time of bind are > copied by value. If you need a reference use the boost::ref wrapper. > Expanding your example: > void DClient::resolveHandler( const boost::system::error_code& error, > int value ); > ... > int special_number = 42; > > hostResolver->async_resolve( *nameQuery, > boost::bind(&DClient::resolveHandler, this, > boost::asio::placeholders::error, > special_number ) ); > > ------------------- > > In the above, when the handler is invoked it will pass the error value > (via the placeholder) and the value of special_number during the bind... > which is 42. If I understand you correctly, then in the following code: void DClient::resolveHandler(int param) { .... } boost::function<void (int)> functionPtr; int tempInt = 12; myHandler = boost::bind(&DClient::resolveHandler, this, tempInt); myHandler(67); The final line will call resolveHandler with a parameter of 12, rather than 67? Surely that can't be right?

3 4

Re: [Boost-users] [msm] EUML submachine without macro
by ray＠maya-studio.com 29 Jun '11

29 Jun '11

Christophe Henry wrote: > you're just a little too fast as I added the example in the 1.47 release. In > the meantime, the submachine example can be found there: > http://svn.boost.org/svn/boost/trunk/libs/msm/doc/HTML/examples/CompositeTu… Thank you for quick reply! Works like a charm. Kudos for your excellent library! Best Regards, Ray

3 4

Re: [Boost-users] Boost-users Digest, Vol 2767, Issue 2
by Marat Abrarov 29 Jun '11

29 Jun '11

> I am grateful for the link, Marat, although I can't do exactly what you are doing. We are constrained > as to what tools we can use with our embedded target's Linux. Oh, I missed this. > I've got to use the proper gcc, gmake, > and libs. It's going to be painful enough to get the target's boost libraries upgraded because some of > them will cause API breakages. Boost.Asio is header only library. I often upgrade Boost with most recent Boost.Asio version by overwriting files. Regards, Marat Abrarov.

2 1

BGL Traveling Salesman Approximation
by David Doria 29 Jun '11

29 Jun '11

Using the example here: http://www.boost.org/doc/libs/1_44_0/libs/graph/test/metric_tsp_approx.cpp I constructed this path: http://homepages.rpi.edu/~doriad/Upload/contour.png on these points: http://homepages.rpi.edu/~doriad/Upload/points.png Is this the best I can hope to do with this approximate algorithm? (There is "clearly" (to a human) a much shorter path without all of the overlapping edges on the lower part of the left side of the rectangle). The graph.txt file I used is here: http://homepages.rpi.edu/~doriad/Upload/graph.txt and the solution it gave was: 0 1 2 3 5 7 9 11 13 15 16 18 20 21 23 25 26 28 30 32 35 37 40 39 38 36 34 33 31 29 27 24 22 4 6 8 10 12 14 17 19 0 Is there anyway I can "fix" this solution? Thanks, David

2 1

filesystem3::rename error codes
by Andreas Haferburg 29 Jun '11

29 Jun '11

Greetings, I was wondering how to handle errors of filesystem3::rename(). I'd like to use the system::error_code& overload, but I can't seem to find out how to use it. The documentation at [1] points to POSIX rename(), so I'm guessing the error_code somehow corresponds to the POSIX rename() return value. I'm not a native speaker, but I find the documentation pretty vague here. The tutorial section [2] doesn't help either. For example if the target file doesn't exist, the error_code.value() is 3. But I'm not supposed to use the constant 3 in the code, am I? Are there any constants defined that correspond to these error code values? What's the proper way to do the following: TEST( BoostFilesystem3Test, testRenameFails ) { boost::system::error_code ec; boost::filesystem3::rename("c:/does/not/exist", "asdf", ec); ASSERT_EQ(3, ec.value()); } I'm going to use QFile for now, but only because their documentation is (slightly) better. I would like to use boost for as much as possible, so an explanation or pointers to better documentation would be much appreciated. Thank you. Best regards Andreas P.S. I accidentally sent this mail to the boost@ list before, sorry about that. [1] http://www.boost.org/doc/libs/1_46_1/libs/filesystem/v3/doc/reference.html#… [2] http://www.boost.org/doc/libs/1_46_1/libs/filesystem/v3/doc/tutorial.html#E…

1 0

Using Boost::Bind
by David McCallum 29 Jun '11

29 Jun '11

I'm having some difficulty passing parameters to boost::bind. The approach taken by ASIO is to pass a placeholder to bind, like so: hostResolver->async_resolve(*nameQuery, boost::bind(&DClient::resolveHandler, this, boost::asio::placeholders::error, boost::asio::placeholders::iterator)); This works fine if I want to use an error or iterator as a parameter. But what if I want to use a primitive variable, or a pointer to one of my classes? Do I need to make a placeholder for each type of parameter I want to use? I looked in ASIO's placeholders.hpp file, and I must confess I can't make sense of it at all! Any help here is appreciated.

3 3

threading/io service query
by Sean Farrow 28 Jun '11

28 Jun '11

Hi Fokes: I'm currently working on a piece of software that is using or going to use boost. There are two classes that use the boost threading components: The first MidiWidget allows the opening of midi input/output ports, I plan to associate two threads, one to receive data and one to output the data. The second is where the problem could occur, MidiSurface simulates a midi control surface, which processes midi messages. Given that there could be multiple surfaces in the program-determined by connections to the pc, and that each surface may need to process a different set of messages which are defined in an xml file, what do people consider is the best approach to firing up a thread when ever a surface is created. I'd ideally like each surfaces thread be able to examine the data contained within it's owning surface without copying the whole class. Is it better to have the thread function as a virtual function in the surface class or a separate function that takes the surface, or a reference to the surface as a parameter. Also I've got a thread safe queue, but was wondering whether there is now one included with boost-or tr1. I'm using boost 1.46.1 and visual studio 2010 sp1 with the latest windows sdk. Any help/thoughts appreciateds. Regards Sean.

1 0

BGL: Copying graphs and edge_descriptors confusion
by David Doria 28 Jun '11

28 Jun '11

I have a graph, 'inputGraph' with a bool in its bundled edge properties. If I copy that graph with: Graph g = inputGraph; there seems to be some odd behavior with the edge_descriptors. If you take a look at this example: http://programmingexamples.net/index.php?title=CPP/Boost/BGL/CopyAGraph I set one of the edges to visible=0 and then output all of the edge visibilities using the iterator from boost::edges(g) - this works properly, the edge that I set to visible=0 indeeds shows up as such. However, if I instead use the iterator from boost::edges(inputGraph) , all of the edges appear as visible! Shouldn't the edge_descriptors be identical since the graph g is an exact copy of inputGraph? Can anyone explain why the last two loops in that example do not produce the same output? Thanks, David

3 5