June 2012 - Boost-users - lists.preview.boost.org

[serialization] mingw linker problems
by Ken 23 Jun '12

23 Jun '12

Hi, We have an application comprised of many shared libraries. It is built on Windows using mingw (gcc 4.4) and on Mac OS X using gcc under the Qt 4.7.0 framework. The application has been based on boost 1_41. We are now attempting to port to boost 1_49. We use boost.serialization and other boost libraries as dlls. On Mac OS X the upgrade was mostly trouble free, but with mingw we have encountered problems with all of our libraries that use boost.serialization. Basically, the inclusion of boost.serialization code appears to alter the default symbol visibility of the dlls generated by mingw. Each of our libraries consists of a declaration of a class, with public functions, and zero or more free functions. If the library implementation includes boost.archive or boost.serialization the public functions of the class, and any free functions declared, are missing from the import library, though visible in the dll. Each of our libraries has a test executable, and with many it is a trivial task to comment out the use of boost.serialization in the library and test code. When we do that, there are no linker problems. As soon as we reintroduce functions that use boost.serialization into the library, the test function fails to find the public functions of the library class. Please can anyone explain this? How does the use of boost.serialization change the global behaviour of mingw gcc in generating dll import libraries? TIA Ken Appleby

2 2

[phoenix] passing values
by Karsten Ahnert 23 Jun '12

23 Jun '12

Hi, I figured out that argument deduction in phoenix only works with references, so the following will not compile: std::cout << arg1( 1.0 ) << std::endl; Are there any alternatives to use use phoenix directly with values?

2 1

[mpl] Documentation suggestion
by Jeremiah Willcock 22 Jun '12

22 Jun '12

I have noticed that several MPL algorithms do not work when applied to views unless an output iterator (inserter) is passed in explicitly. The error messages when you hit one of those cases are very non-obvious, however. It might be nice to have something in the documentation that talks about this issue and the workarounds for it. Here is an example of something that breaks: #include <boost/mpl/vector_c.hpp> #include <boost/mpl/joint_view.hpp> #include <boost/mpl/sort.hpp> #include <boost/mpl/print.hpp> typedef boost::mpl::vector_c<int, 2, 0> va; typedef boost::mpl::vector_c<int, 1, 3, -1, 5> vb; typedef boost::mpl::joint_view<va, vb> vc; typedef boost::mpl::sort<vc>::type vd; The first important part of the error message with GCC 4.7.0 in C++11 mode is: .../boost-svn/boost/mpl/clear.hpp: In instantiation of ‘struct boost::mpl::clear<boost::mpl::joint_view<boost::mpl::vector_c<int, 2l, 0l>, boost::mpl::vector_c<int, 1l, 3l, -1l, 5l> > >’: .../boost-svn/boost/mpl/sort.hpp:23:1: required from ‘struct boost::mpl::sort<boost::mpl::joint_view<boost::mpl::vector_c<int, 2l, 0l>, boost::mpl::vector_c<int, 1l, 3l, -1l, 5l> > >’ mpl_test.cpp:9:29: required from here .../boost-svn/boost/mpl/clear.hpp:28:8: error: invalid use of incomplete type ‘struct boost::mpl::clear_impl<boost::mpl::aux::joint_view_tag>::apply<boost::mpl::joint_view<boost::mpl::vector_c<int, 2l, 0l>, boost::mpl::vector_c<int, 1l, 3l, -1l, 5l> > >’ which seems to indicate that the default inserter needs a "real" container as input to the algorithm in order to know what container to use for the result. Since this limitation is probably unavoidable, some kind of note in the documentation might be useful. Thanks. -- Jeremiah Willcock

2 1

Arguments types for object created with boost::bind()
by niXman 21 Jun '12

21 Jun '12

Hello! I need to get the types list for operator() for result type of boost::bind() Example: void func(int, long) {} auto o = boost::bind(&func, _1, _2); o( ??? ); // what types of arguments? Thanks. -- Regards, niXman ___________________________________________________ Dual-target(32 & 64 bit) MinGW compilers for 32 and 64 bit Windows: http://sourceforge.net/projects/mingwbuilds/

3 3

boost.log on Windows with MingGW
by Konstantin Ribel 21 Jun '12

21 Jun '12

1 0

[range] indirected requirements for custom smart pointer
by Jeff Flinn 21 Jun '12

21 Jun '12

Using boost::adaptors::indirected with a custom templated smart pointer failed compilation due to missing MyCustomSmartPointer::element_type typedef. Indeed supplying this typedef, all compiles and tests run. Is this the canonical approach to making a compatible pointer type usable with range? Is this specific to range only? Will this make my custom smart pointer 'more' compatible with other boost/std facilities? Thanks, Jeff

2 1

Re: [Boost-users] Boost Program Options: Any way to have variable number of "objects" each with parameters?
by Charles Mills 21 Jun '12

21 Jun '12

Here is what I have come up with. Any feedback? I first thought of something like this: CircleRadiuses= 4 7 3 . It's okay for objects with only one value (like Circles) but clumsy and unintuitive for objects with more than one parameter like RectangleHeights= 5 3 7 . RectangleWidths= 10 4 16 . Although I would like to support "any number" of objects, the reality is that a likely number is three or four, so restricting it to a maximum of nine should not be a problem. So here is my plan. Something like [Rectangle1] Height=5 Width=10 [Rectangle2] Height=3 Width=4 and so forth through Rectangle9. Same sort of deal for Circle1 through 9. (I think counting from 1 is more user-friendly for people who are not C programmers.) So, any feedback? Any better way to solve this problem? Charles From: Charles Mills [mailto:charlesm@mcn.org] Sent: Wednesday, June 20, 2012 6:20 PM To: 'boost-users(a)lists.boost.org' Subject: Boost Program Options: Any way to have variable number of "objects" each with parameters? I really like the Boost Program Options library. I've used it for one program now with great effect and I'd like to use it for another, but can't figure out how to structure what I need to do so that Boost PO could handle it. I am solely interested in the ".ini file" for this - command line is irrelevant. The problem is "how does a program allow the specification of a variable number of 'objects' each of which has parameters of its own?" Here is a hypothetical program that exemplifies the problem. Suppose a program were going to create multiple geometric shapes of some sort. An ideal (unsupported by PO) .ini file might look like rectangles=3 rectangle[0].width=5 rectangle[0].height=10 rectangle[1].width=3 etc. circles=2 circle[0].diameter=4 etc. Can anyone suggest a way to accomplish this with Boost PO? I'm real flexible on what the .ini file might look like so long as it solves the problem and is reasonably explicable to a user. (No need for it to look almost like the above.) Thanks, Charles

1 0

Using Boost ICL for range matching
by Rajalakshmi Iyer 21 Jun '12

21 Jun '12

Hello, I want to know if I can use the boost::icl::interval_set to store a set of integer based intervals like: [100, 200], [50, 600] etc and then find out whether a number like 150 matches any of these intervals? I would greatly appreciate if someone can point me to an example for this kind of operation. Thanks, Raj -- Twitter: @Blismedia <http://twitter.com/#%21/blismobile> BlisMedia 32 Percy Street, London W1T 2DE www.blismedia.com <http://www.blismobile.com/> <http://www.blismobile.com/> [image: Follow on Twitter] <http://twitter.com/#%21/blismobile>[image: Blis Website] <http://www.blismobile.com/> This communication is from Blis Media Ltd, a company registered in England and Wales with registered number 06455773. Its registered office is 32 Percy Street, London W1T 2DE, United Kingdom. This communication contains information that is confidential and may also be privileged. It is for the exclusive use of the intended recipient(s). If you are not the intended recipient(s), please (1) notify info(a)blismedia.com<info(a)blismobile.com> by forwarding this email and delete all copies from your system and (2) note that disclosure, distribution, copying or use of this communication is strictly prohibited. Email communications cannot be guaranteed to be secure or free from error or viruses. All emails sent to or from a Blismobile email account are securely archived and stored by an external supplier. This email does not constitute a contractual agreement; such agreements are in specified contractual or Insertion Order (IO) form only and exclusively contain all the terms to which Blis Mobile Ltd will be bound. To the extent permitted by law, Blis Mobile Ltd does not accept any liability for use of or reliance on the contents of this email by any person save by the intended recipient(s) to the extent agreed in a contract or Insertion Order. Opinions, conclusions and other information in this email which have not been delivered by way of the business of Breeze Tech (UK) Ltd are neither given nor endorsed by it.

2 3

Re: [Boost-users] the problem about boost trim function in multi-thread
by Hu Jing 21 Jun '12

21 Jun '12

>>That should be really easy to test. >>How does your test code perform if you change it to this: >> >>void *TrimNString(void *arg) { >> >> string base ="fdsffdsafdsa"; >> locale loc = locale(); >> for(int i = 0; i != 50000000;i++) >> { >> string str = base; >> trim(str, loc); >> } >> return 0; >> } >> Note also that trim calls the locale to look for whitespace, which is more flexible than your test for " \r\n\t". >> That's probably a significant performance difference, too. I have try you method, But it doesn't work. I don't think that boost trim using the locale cause the problem. In the single thread envoriment, the program has the same running time as the my test. so it does't change the program complexity.

1 0

[preprocessor] Identity function for token concatenation
by Jeremiah Willcock 20 Jun '12

20 Jun '12

Boost.Preprocessor has a macro BOOST_PP_INTERCEPT to eat a numeric value that it is token-concatenated onto. Is there a similar macro that just returns the value? I.e., some BOOST_PP_EMPTY_FOR_CONCAT such that BOOST_PP_EMPTY_FOR_CONCAT ## 3 turns into 3? As with BOOST_PP_INTERCEPT, it only needs to work for small integer values. Is there some other technique I can use for this? I am using it to generate std::get<> invocations in BOOST_PP_ENUM_BINARY_PARAMS. Thank you for your help. -- Jeremiah Willcock

3 8