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

filesystem hidden files
by Kraus Philipp 28 Jun '11

28 Jun '11

Hello, I use the Boost Filesystem to browser the data within a directory. On unix systems there are file with a dot on begin, the hidden files. Is there a method for checking the path variable if the file is hidden) like path p("~/.myhiddenfile"); if (is_regular_file(p) && is_hidden(p)) I would like to exclude hidden files of my browsing list. Thanks Phil

1 0

making boost build generic library names (without -mgw45-mt-1_46_1)?
by Ákos Maróy 28 Jun '11

28 Jun '11

Hi, I wonder how can one make the boost build process create libraries with generic boost library names (like libboost_math_c99f.a) on Windows using mingw / gcc, as opposed to library names like libboost_math_c99f-mgw45-mt-1_46_1.a? Although I understand this makes sense in some contexts, a lot of times this is not convenient. For example, I'd have to update my dependencies every time I change the boost version, or the underlying gcc version changes. moreover, on other platforms (Linux, MacOS), the libraries built have the generic name (libboost_math_c99f.a) How to achieve the same effect on Windows? Akos

2 2

Does Boost use a platform-configurable [[ noreturn]] attribute ?
by John M. Dlugosz 27 Jun '11

27 Jun '11

Does the Boost 1.46 library have a macro or something that I can use to decorate a function as "noreturn" in a platform independent manner? Is there documentation on platform-independence tools that I might use in general, as Boost itself uses them? —John

1 0

Boost.Format to a string
by John M. Dlugosz 27 Jun '11

27 Jun '11

I brought this up a while back, and I ran into it again yesterday. Using boost::format to produce a string, as opposed to sending it directly to cout, is awkward. boost::format ("blah blah blah") % exp1 % exp2 %exp3 is fine in and of itself. But then what? You need to wrap the whole thing in extra parens to use str: std::string s= (boost::format ("blah blah blah") % exp1 % exp2 %exp3).str(); WHY NOT allow implicit conversion to string? Or, some other mechanism that doesn't complicate the expression and bury the interesting part? —John

2 2

[msm] how to write a generic class for all state machine back-end.
by Jerry Chen 27 Jun '11

27 Jun '11

In our project, we need a generic interface to refer to different state machine (back-end). For example, in the following codes, we want to use SMBase* to refer to any concrete state machine. SMBase should have similar interface as state machine. struct User { A(SMBase* s) : sm(s) (); SMBase* sm; ... // User can use sm->start(), sm->process_event() etc } struct SMBase { virtual void start() = 0; ... // how to design the rest ??? } // a wrapper implement SMBase template < typename SM > class SMWrapper : public SMBase{ SMWrapper(SM & s): sm(s){}; virtual void start() { sm.start(); } ... SM sm; } main() { typedef boost::msm::back::state_machine SM1; typedef boost::msm::back::state_machine SM2; User a1(new SMWrapper(SM1()) ); // Wrap the concrete state machine and pass it to User. User a2(new SMWrapper(SM2()) ); // Pass a different state machine. ... } The problem of designing SMBase is the function "process_event". We want it to to be virtual function so SMWrapper can implement it. But it also needs to be a template according to state machine interface. But we cannot have a virtual template function. How can we achieve/design this generic state machine interface? Many thanks. Jerry

4 4

Re: [Boost-users] [boost] [review][assign] Formal review of Assign v2 ends today
by Lars Viklund 27 Jun '11

27 Jun '11

On Fri, Jun 24, 2011 at 11:56:14PM +0100, John Bytheway wrote: > The formal review of the Assign v2 library upgrade by Erwann Rogard is > scheduled to end today. > > Despite the end of the review, please feel free to discuss the library > further, make any final comments, or even submit a late review. I will > certainly not declare the review request before next Monday, and will > incorporate any comments made before then. Hopefully this is not too late to be counted as a proper mini-review. The test suite for the library fails miserably on VisualAge C++ 11.1, as can be seen in the logs at [1], which doesn't bode well for a newly written library aimed for inclusion in Boost. The test suite fails similarly on SunStudio 12.1 on Solaris 10, with the results seen in [2]. As a comparison, the tests for assign_v1 passes with flying colours with VisualAge. Having the replacement library support way fewer platforms in Boost sounds very strange to me, and is alone grounds for not including the library. (I was not able to complete building the v1 tests on the Solaris box I had, as the linker seems slightly broken, something that doesn't affect the v2 tests, as none of them can compile.) The performance metrics seem to lack a very important metric - namely compile-time costs in memory usage and compile time increases. The documentation mentions compile-time costs briefly in lamenting that the test suite takes several minutes to build, which isn't horribly encouraging. A library that is intended to be easily dropped into any TU you might need it in should be very cheap to use. I currently avoid using Karma/Qi even for things where they would simplify things, solely due to the cost of including them. Regarding the code, I'm a bit concerned about the namespace layout. The current setup will cause any users of 1.0 that also happen to include 2.0 indirectly to end up with namespaces 'v2' and 'ref' injected wherever they previously did 'using namespace boost::assign;'. As for naming of the functions and types, I find them both opaque and overly generic. 'put()' has a strong risk of being used already in user code, and things like 'csv()' are utterly meaningless without having read the documentation, especially with regard to the non-type template argument. The use of asdf_aux namespaces seems a bit off, considering that the canonical naming for implementation detail is just that, 'detail'. This library feels misguided, trying to be something the name doesn't indicate it should be. I found assign_v1 to be an excellent library for quick and dirty initialization of sequences. The whole transformation pipeline that v2 introduces feels out of place and most probably belongs in a quite different library. The documentation even mentions Boost.Range's sequence adaptors - which makes me wonder if they're not better off integrated with Range? As for the tutorial, which is the closest thing to examples I find, they seem overly terse and seems to assume that the concepts used in the library are already known. My vote for this library is: NO. [1] http://www.acc.umu.se/~zao/assign_v2/vacpp-test-20110627-1014.txt [2] http://www.acc.umu.se/~zao/assign_v2/suncc-test-20110627-1044.txt -- Lars Viklund | zao(a)acc.umu.se

2 2

Re: [Boost-users] Directory Iterator / Directory "access denied" exception
by Tan, Tom (Shanghai) 27 Jun '11

27 Jun '11

I have filed a Ticket #5403 and applied a patch here some time ago: https://svn.boost.org/trac/boost/ticket/5403 Hope it works for you. -----Original Message----- Date: Fri, 24 Jun 2011 01:48:05 -0700 (PDT) From: Kiah08 <tonycaver2007(a)gmail.com> Subject: [Boost-users] Directory Iterator / Directory "access denied" exception Does anyone know of a way to handle exceptions when a directory is inaccessible? I would like to be able to resume the directory scan after the inaccessible directory. Please give an in depth example if possible. Using Boost 1.46.

1 0

GCC 4.6.0 rejects Boost Intrusive header
by Brian Wood 26 Jun '11

26 Jun '11

GCC 4.6.0 rejects the *Boost* intrusive header list.hpp. Here's the command line: g++ -std=c++0x -O2 -s -W -Wall -I /home/brian/boost_1_46_1 -c -o jaj.o jaj.cc and the output: In file included from jaj.cc:1:0: /home/brian/boost_1_46_1/*boost*/intrusive/list.hpp: In static member function =91static *boost*::intrusive::list_impl<Config>& *boost*::intrusive::list_impl<Config>::priv_container_from_end_iterator(const const_iterator&)=92: /home/brian/boost_1_46_1/*boost*/intrusive/list.hpp:1291:89: sorry, unimplemented: unexpected ast of kind field_decl /home/brian/boost_1_46_1/*boost*/intrusive/list.hpp:1291: confused by earlier errors, bailing out Preprocessed source stored into /tmp/ccRWz6ZT.out file, please attach this to your bugreport. The .cc file is one line -- an include of list.hpp. Here's the function that it is complaining about: private: static list_impl &priv_container_from_end_iterator(const const_iterator &end_iterator) { root_plus_size *r = detail::parent_from_member<root_plus_size, node> ( detail::boost_intrusive_get_pointer(end_iterator.pointed_node()), &root_plus_size::root_); data_t *d = detail::parent_from_member<data_t, root_plus_size> ( r, &data_t::root_plus_size_); list_impl *s = detail::parent_from_member<list_impl, data_t>(d, &list_impl::data_); return *s; } 1291 is the line that starts with "list_impl *s = " I'd like to be able to begin using GCC 4.6, but this is preventing me from doing so. Would someone with GCC 4.7 or 4.6.1 test this and report the results? TIA. -- Brian Wood Ebenezer Enterprises http://webEbenezer.net <http://webebenezer.net/> Trust in the L-RD and lean not on your own understanding. In all your ways acknowledge Him and He will direct your paths. Proverbs 3:5,6

2 1

[Spirit] Problem while parsing a quoted string
by Sven Steckmann 25 Jun '11

25 Jun '11

Dear list, at the moment I am writing a grammar description in which I want to used quoted string. Up to now I hardcoded the quotes to '"', but now, I want to enable this rule working with a parameter. Maybe I am missing somethin, I hope one can give me good hint: This is the rule definition (for parsing QStrings, I took the code from the example): quotedQString %= lexeme[lit(_r1) >> +(char_ - lit(_r1)) >> lit(_r1)]; qi::rule quotedQString; I am uisng the quotedQString as follows: nextRule = quotedQString('"') [ anAction(_var,_1)]; This does not work as expected, if I parse a string looking like this "String" I always got the quotes included. My old version: quotedQString %= lexeme[lit('"') >> +(char_ - lit('"')) >> lit('"')]; works quite as expected and parses everything without quotes. Kind regards, Sven Steckmann --

1 0

[1.47.0] exception build problem
by Orhun Birsoy 25 Jun '11

25 Jun '11

When I build boost 1.47.0 beta 1 with the following command bjam --stagedir=vc10-x86-stage --layout=tagged toolset=msvc-10.0 variant=debug link=shared threading=multi runtime-link=shared address-model=32 --with-exception I get the following output. [snip] msvc.link.dll bin.v2\libs\exception\build\msvc-10.0\debug\address-model-32\threading-multi\boost_exception-mt-gd.dll msvc.manifest.dll bin.v2\libs\exception\build\msvc-10.0\debug\address-model-32\threading-multi\boost_exception-mt-gd.dll common.copy vc10-x86-stage\lib\boost_exception-mt-gd.dll bin.v2\libs\exception\build\msvc-10.0\debug\address-model-32\threading-multi\boost_exception-mt-gd.dll 1 file(s) copied. common.copy vc10-x86-stage\lib\boost_exception-mt-gd.lib The system cannot find the file specified. copy /b "bin.v2\libs\exception\build\msvc-10.0\debug\address-model-32\threading-multi\boost_exception-mt-gd.lib" + this-file-does-not-exist-A698EE7806899E69 "vc10-x86-stage\lib\boost_exception-mt-gd.lib" ...failed common.copy vc10-x86-stage\lib\boost_exception-mt-gd.lib... ...failed updating 1 target... ...updated 15 targets... It looks like boost_exception-mt-gd.dll doesn't export any symbols. -- Orhun Birsoy

1 0