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

Re: [Boost-users] [boost] Boost.Operators issue: polluting namespace of caller through argument-dependent lookup
by John M. Dlugosz 21 Jan '16

21 Jan '16

On 1/20/2016 10:41 AM, Daniel Frey wrote: > In a way it makes library collections like Boost harder and I still think you should use > qualified lookup in your specific case, No, I specifically don't want to qualify every use, as explained originally. The code needs compatibility with some configurations that require the tr1 namespace and others that prohibit it. Qualifying every reference is bad; rather, import them in one place. You are basically saying that `using N::foo;` is never right because someone may happen to use the name foo in an unrelated library and ADL is a misfeature? I think that doesn't scale. It should be SOP to isolate classes that are meant to be used as base classes by library consumers, and treat unqualified lookup of names pulled in to the local scope via `using` as something that is a useful feature that ought to work. > but as I already said in another mail I’m willing to fix it. I just need to figure out > where to commit and merge stuff as the last time I worked on it we were still using svn > and the branching model was different. I hope to be able to find the time somewhere in > the next days. Regards, Daniel Looking forward to it in 1.62 or whatever. Thanks! —John

1 0

[locale] boost::locale::format number formatting on OS X and Windows
by Michael Smith 21 Jan '16

21 Jan '16

Using Boost.Locale from Boost 1.60.0 (installed via homebrew, so it uses icu) on Mac OS X 10.11.2, I seem to be getting unexpected behavior for some of the boost::format number formatting. A short example program #include <boost/locale.hpp> int main(int argc, char** argv) { boost::locale::format fmt("{1,number} should be 1,25."); fmt % 1.25; boost::locale::generator gen; std::cout << fmt.str(gen("fr-fr.UTF-8")) << std::endl; return 0; } That produces 1,25 on Linux, and 1.25 on Windows (using mingw-w64) and Mac. Any idea why they'd represent it differently?

1 0

Re: [Boost-users] Using Coroutines in the Visitor Pattern (use-case: BGL event visitor)
by alex 20 Jan '16

20 Jan '16

>-----Original Message----- >From: Boost-users [mailto:boost-users-bounces@lists.boost.org] On Behalf Of >Daniel Hofmann >Sent: 17 November 2015 15:55 >To: boost-users(a)lists.boost.org >Subject: [Boost-users] Using Coroutines in the Visitor Pattern (use-case: BGL >event visitor) > >(Disclaimer: this is not only specific to BGL, I can think of similar >use-cases where the visitor pattern is used --- I will still explain it >in the BGL context, as it is probably easier to understand) > > >Suppose you have a BGL graph and want to implement s-t shortest path >queries. It is possible to do this by using a bi-directional approach, >that is starting a Dijkstra search [1] from the start and starting a >additional Dijkstra search from the target. As soon as they "meet in the >middle", you can unpack the two sub-paths. > >[snip] > >It then came to my mind that I could use Coroutines for cooperative >multitasking, without the downsides that come with threading. > >My idea was to invert the callbacks using Coroutines as it is shown in >the SAX parsing example [3], in order to get a lazy range of vertices >each visitor examines. I then could simply walk the two ranges, >terminating as soon as there is the same element in both. Jeremiah Willcock gave a good description of how this could work: http://lists.boost.org/Archives/boost/2012/07/195300.php The way he describes it is really using coroutines instead of the visitor pattern. > > >But due to way the dijkstra_shortest_path function adds an additional >level of indirection I failed to implement this, and I have strong >doubts that it is possible with Coroutines at all. I> >I'm mainly struggling with wrapping my head around stopping and resuming >the dijkstra_shortest_path function, when all I can modify is the >visitor it takes. That is, the reason I can not make it work is that I >can not launch two dijkstra_shortest_path functions asynchronously with >just Coroutines. > I think there are two options: 1. Let your visitor throw an exception to get out of the dijkstra shortest path function; use some version of the function that skips initialization to resume again (the existing dijkstra_shortest_path_no_init is not good enough, so you would have to make one). 2. Reimplement the dijkstra function as a coroutine. I have implemented both, but have only really used option 1; whereas option 2 remained a toy experiment. In both cases the tricky part was to properly initialize and keep track of(and not re-initialize on resume) all parameters that are input to the dijkstra function, i.e. color_map, queue, distance_map, etc. https://github.com/ahhz/resumable_dijkstra I used the lightweight stackless coroutine of Boost ASIO. http://www.boost.org/doc/libs/1_59_0/doc/html/boost_asio/reference/coroutine .html >The control flow I need would probably look like this: start the first >Dijkstra search, on its first vertex it visits, "jump back out" and >start the second Dijkstra visitor. From there on ping-pong between the >two visitors exclusively. And I'm not sure this is possible with the >Coroutine approach. > It depends how much you are willing to re-implement. It is possible in principle >I appreciate any tips in this regard, even if it's a "just use your 15 >lines of threading code instead of trying to be fancy" :) > I think it would be great to have a fancy solution for this in BGL.

3 14

boost serialization
by Виктория Доможакова 20 Jan '16

20 Jan '16

I use boost serialization in my project. I I conducted f an experiment: 1) I removed one line with the data in the boost archive file. 2) I started my project. 3) All variables initialized in the constructor. 4) Value of variables loaded from boost archive in order . 5) When queue comes to removing values , I get exeption (I handle exeption with try-catch ) and file didn't read to end and close. 6) But this variable is equal to the value next after removal . My question: why this variable is equal to the value next after removal . if this variable has value from constructor than it is more logic. -- Виктория Доможакова

1 0

[gsoc16] Attn: We need 2016 mentors and project ideas for Boost!
by Niall Douglas 20 Jan '16

20 Jan '16

Dear Boost community, We are now within one month of the beginning of Google Summer of Code 2016! Last year's GSoC was a particularly momentus one as we witnessed the approval by the community for Boost.Hana to enter Boost - congratulations to Louis and his mentor Joel for bringing to Boost a very exciting new C++ library which pushes C++ compilers to their limits. We have also seen greatly improved retention of GSoC students in the community thanks to the improved student selection procedures introduced last year, and the fresh blood brought into Boost has been very evident on the mailing list this year. As part of filling in our application for 2016, we must supply to Google a list of potential GSoC mentors and potential GSoC projects for summer 2016 in order to be approved as an organisation this year. Therefore a lengthy page of awesome project ideas at https://svn.boost.org/trac/boost/wiki/SoC2016 before *February 8th* would be an enormous help to getting us approved again this year and bringing to Boost another US$35,000 of Google funding. If you think yourself able to mentor a student doing some work on Boost this summer, *please* consider adding a description of the proposed work item and your name to the list. Last year we had more slots awarded to us by Google than we had good combinations of student and mentor, so we had to return slots which I believe has not happened here at Boost GSoC in at least a decade. That means that your project ideas have a better chance than ever of receiving funding if you can find a good student, and that is best done by getting project ideas floated *early*. If you want to know more about mentoring a Google Summer of Code funded student work before you nominate yourself, please feel free to ask on the main Boost developers mailing list boost(a)lists.boost.org or via gsoc-admin(a)boost.org. Thank you in advance for your time. --- Boost C++ Libraries Google Summer of Code 2016 admin https://svn.boost.org/trac/boost/wiki/SoC2016

2 1

Using Coroutines in the Visitor Pattern (use-case: BGL event visitor)
by Daniel Hofmann 19 Jan '16

19 Jan '16

(Disclaimer: this is not only specific to BGL, I can think of similar use-cases where the visitor pattern is used --- I will still explain it in the BGL context, as it is probably easier to understand) Suppose you have a BGL graph and want to implement s-t shortest path queries. It is possible to do this by using a bi-directional approach, that is starting a Dijkstra search [1] from the start and starting a additional Dijkstra search from the target. As soon as they "meet in the middle", you can unpack the two sub-paths. In terms of the BGL this would roughly look like the following (simplified, with synchronization and visitor omitted): Vertex middle; async(dijkstra_shortest_paths(g, start, visitor(v{middle}))); async(dijkstra_shortest_paths(g, target, visitor(v{middle}))); where the visitors single-step all vertices in their examine_vertex callback [2] and check if they met in the middle. While this is quite easy to implement with threading, is has its downsides, mainly in synchronization for this ping-pong play between the two visitors, resulting in contention on the synchronization locks. But also the overhead of using threads might be too high for smaller graphs. It then came to my mind that I could use Coroutines for cooperative multitasking, without the downsides that come with threading. My idea was to invert the callbacks using Coroutines as it is shown in the SAX parsing example [3], in order to get a lazy range of vertices each visitor examines. I then could simply walk the two ranges, terminating as soon as there is the same element in both. But due to way the dijkstra_shortest_path function adds an additional level of indirection I failed to implement this, and I have strong doubts that it is possible with Coroutines at all. I'm mainly struggling with wrapping my head around stopping and resuming the dijkstra_shortest_path function, when all I can modify is the visitor it takes. That is, the reason I can not make it work is that I can not launch two dijkstra_shortest_path functions asynchronously with just Coroutines. The control flow I need would probably look like this: start the first Dijkstra search, on its first vertex it visits, "jump back out" and start the second Dijkstra visitor. From there on ping-pong between the two visitors exclusively. And I'm not sure this is possible with the Coroutine approach. I appreciate any tips in this regard, even if it's a "just use your 15 lines of threading code instead of trying to be fancy" :) Cheers, Daniel J H [1] http://www.boost.org/doc/libs/1_59_0/libs/graph/doc/dijkstra_shortest_paths… [2] http://www.boost.org/doc/libs/1_59_0/libs/graph/doc/DijkstraVisitor.html [3] http://www.boost.org/doc/libs/1_59_0/libs/coroutine/doc/html/coroutine/moti…

2 4

Re: [Boost-users] boost.dll winapi compile error
by Elod G 18 Jan '16

18 Jan '16

I was missing the latest winapi module, duh. ElodG On 1/18/2016 19:00, boost-users-request(a)lists.boost.org wrote: > Message: 1 > Date: Sun, 17 Jan 2016 23:12:32 +0200 > From: Elod G <gyelod(a)gmail.com> > To: boost-users(a)lists.boost.org > Subject: [Boost-users] boost.dll winapi compile error > Message-ID: <569C03BF.9030305(a)gmail.com> > Content-Type: text/plain; charset=utf-8 > > Hello All, > > I cloned origin/master of https://github.com/boostorg/dll and copied > include/ to the 1.60 release of boost to give boost.DLL a go. > > I am getting the following compile error when including: "boost/dll.hpp" > boost\dll\detail\windows\shared_library_impl.hpp:140: error: C2039: > 'get_proc_address' : is not a member of 'boost::detail::winapi' > > I am using VS2013. > > What am I missing? > > -- > ElodG

1 0

boost.dll winapi compile error
by Elod G 17 Jan '16

17 Jan '16

Hello All, I cloned origin/master of https://github.com/boostorg/dll and copied include/ to the 1.60 release of boost to give boost.DLL a go. I am getting the following compile error when including: "boost/dll.hpp" boost\dll\detail\windows\shared_library_impl.hpp:140: error: C2039: 'get_proc_address' : is not a member of 'boost::detail::winapi' I am using VS2013. What am I missing? -- ElodG

1 0

multiprecision thread safety
by Daniel Brake 14 Jan '16

14 Jan '16

Hi Boost, I have a couple of questions regarding thread safety and Boost.Multiprecision. I am writing a library, and would like to be able to use mpfr_float (packed into my own complex class) in differing variable precisions on multiple threads simultaneously. 1. Is this supported? I looked in documentation, but didn't find comments to that end. 2. Must I manually call mpfr_free_cache when terminating threads, as recommended by the mpfr documentation? see http://www.mpfr.org/mpfr-current/mpfr.html, section 4.7 memory handling. If so, since Boost.Multiprecision claims to not change the default precision of mpfr_t, do I need to access the particular back end for the thread, and free its cache? Thanks, Daniel Brake danielthebrake(a)gmail.com

2 1

boost 1.55 asio tcp cpp03 chat_server example memory leaks
by Hung Ruo Han 14 Jan '16

14 Jan '16

Hi all i hope someone could give me a clue where to investigate... i'm running the chat_server example from boost http://www.boost.org/doc/libs/1_55_0/doc/html/boost_asio/example/cpp03/chat… on visual studio 2010, windows 10 and i downloaded boost binaries from : http://sourceforge.net/projects/boost/files/boost-binaries/1.55.0/boost_1_5… i used a script to simulate 30 tcp clients, each thread behavior basically: 1. connect to tcp server 2. start a loop 3. send a message to tcp server 4. receive a message from tcp server 5. sleep 6. back to step 2 the strange fact is when i use windows task manager to monitor the memory consumption. The numbers from columns *private working set* and *shared working set* remains "stable" for almost 18 minutes and after that the values of *private working set* starts to increase almost 5 MB by minute. So my doubts are: 1. Does anyone ever seen anything similar before? 2. What could cause this? Regards -- Hung Ruo Han <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campai…> This email has been sent from a virus-free computer protected by Avast. www.avast.com <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campai…> <#DDB4FAA8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

1 0