- Boost - lists.boost.org

[exception] virtual inheritance troubles
by Eric Niebler 26 Feb '10

26 Feb '10

In xpressive, I'm trying to follow the recommended use of Boost.Exception and use virtual inheritance in my exception class, regex_error. I'm also trying to hew as closely as possible to the standard, which says that regex_error should inherit from std::runtime_error. However, this simple use of Boost.Exception does not compile: #include <string> #include <exception> #include <boost/throw_exception.hpp> #include <boost/exception/exception.hpp> struct regex_error : virtual std::runtime_error , virtual boost::exception { explicit regex_error(std::string const& what) : std::runtime_error(what) {} }; int main() { boost::throw_exception(regex_error("what")); } On VC9, I get the following: > C:\boost\org\trunk\boost/exception/exception.hpp(400) : error C2512: 'std::runtime_error::runtime_error' : no appropriate default constructor available The trouble is that std::runtime_error does not have a default constructor, and that boost::throw_exception doesn't simply throw the exception passed to it; it throws an object of type of clone_impl<regex_error>. Because of the virtual inheritance, clone_impl needs to call std::runtime_error's directly, but it doesn't. It can't know to do that in general. What is the recommended practice here? <aside> It actually doesn't seem to be particularly useful in this case to use virtual inheritance because std::runtime_error inherits from std::exception non-virtually. In order to avoid the pitfall described here <http://tinyurl.com/yb2jpns>, it seems *ever* exception type in the hierarchy needs to use virtual inheritance, and that unfortunately is not the case for the standard exception hierarchy. </aside> -- Eric Niebler BoostPro Computing http://www.boostpro.com

4 6

[forward] looking for help to keep MSVC7.1 from ICEing
by Tobias Schwinger 26 Feb '10

26 Feb '10

Hello, I'm trying to make Forward release-ready and I'd really like to support MSVC 7.1 which is currently ICEing (in msc1.cpp, line 2701). According to online research that ICE typically occurs when the ODR is violated and the compiler fails to diagnose it. Although the code should be valid, it's likely to be about name management. The code uses an inheritance chain to add call operators of different arities. IOW the class template for arity N inherits from the class template for arity N-1. A 'using base::operator();' is used to pull in the base class' overloads. A stripped-down version of this pattern compiles. When I take out the 'using' declaration the ICE disappears (and so do the base class' overloads). Ideas anyone? Anyone keen enough to look into it? In case of the latter, I recommend the simpler of the two headers: trunk/boost/functional/lightweight_forward_adapter.hpp [ link to svn: http://tinyurl.com/ydoyoa4 ] Thanks, Tobias

1 0

Boost.Polygon : How to deal with per point attributes
by philippe.lamontagne.2＠ens.etsmtl.ca 26 Feb '10

26 Feb '10

Hi Boost community, I am currently experimenting with the BOOST polygon library and I have a few questions. Question 1: For my specific application the points inside the polygons need to contain attributes in addition to their x and y coordinates. This extra information is required in a per-point basis. I need to compute intersections and differences of polygons. The operations can be performed regardless on the attributes, but it's important that the attributes be present in the polygons resulting from the operations. Does BOOST provide a mechanism for keeping such per point attributes? I followed the example to make my own point and polygon classes work with BOOST, but for what I can see, the polygons operations are performed using GTL's own data types and transferred back to my types, loosing any attributes in the process. I am missing something here? Question 2: I noticed that if I have multiple points located on the same straight line only the initial and final points will be kept in the resulting polygon and that the intermediate points will be discarded. Whereas this behavior is probably preferable in the general case, it causes me a problem if I want to maintain point attributes because the attributes are not guaranteed to be the same all over the straight line. Is there a way to change this behavior? I'm aware that polygon operations may create new points. I can deal with a few empty attributes as long as original points keep their initial attributes. We hope there are some solutions or workarounds because I strongly appreciate the library. Thanks for the excellent work.

2 2

[Boost][Container] Non mandatory template parameter not defaulted
by Mathieu - 26 Feb '10

26 Feb '10

Hi, I was reading the boost.container vector sources, and I can't figure how this vector instantiations can work : boost::container::vector<int> foo; considering that the vector definition looks like : template <typename T, typename A> vector; As the second template parameter A has no default type, I believe it's actually mandatory but it's not the case as this : boost::container::vector<Type> compiles fine. I've been reading through the whole source file but I can't find where the trick is. if anyone can enlighten me? ps : sorry for my bad english..

2 2

[filesystem] 1.40 documentation refers to wrong API version?
by Johan Nilsson 26 Feb '10

26 Feb '10

Hi, the Boost.Filesystem documentation at e.g.: http://www.boost.org/doc/libs/1_40_0/libs/filesystem/doc/reference.html#Ope… seems to refer to more recent version. As a specific example, there are no overloads of remove in 1.40 that accepts an instance of boost::system::error_code (as opposed to what the documentation indicates) - easily verifiable in e.g.: http://svn.boost.org/svn/boost/tags/release/Boost_1_40_0/boost/filesystem/o… Is the actual 1.40 documentation available online somewhere? / Johan

1 0

Re: [boost] review system in place is extremely slow? (was Re: [rfc] rcpp)
by Brian Wood 26 Feb '10

26 Feb '10

Mathias Gaunard writes: >> Currently there are a lot of libraries to review, but no review managers. >> That means that the user community don't want to spend a little bit of >> their time to manage a review. >> >> In addition, the last review didn't had too much of reviewers >> (I'm also concerned by this point) > > So you can see, indeed, that this is fairly concerning. I think this may be due to economic changes. When Boost was getting started the economy was better than today. I think Boost should work on turning traffic on it's pages into advertising revenue. That money could be used to help those who want to be involved in the project. Giving a review manager $200 for 40+ hours of work only makes sense. -- Brian Wood Ebenezer Enterprises http://webEbenezer.net (651) 251-9384

5 4

Re: [boost] review system in place is extremely slow? (was Re: [rfc] rcpp)
by Brian Wood 25 Feb '10

25 Feb '10

On 02/25/2010 11:56 AM, Nevin Liber wrote: > Then comes the "which jobs do we pay for?". If a reviewer > spends 80 hours doing a detailed review, should we pay > them too? What about all the people who tirelessly work > on getting a release out the door every quarter? Why > should they work "for free" if other people are getting paid? I thought about some of these things as well.... I think release managers should be paid as well and more than review managers. For reviewers, I would set up a system that rewarded them also. The review manager might be expected to judge who has given the best and second best reviews. It would add a little excitement to the whole process... seeing who wins the gold and silver medals. > If we don't make enough money from web advertising, you > can't just not pay the review managers. Who will make up > the deficit? Etc., etc. > The way I'd do it is to make offers based on money already collected and not based on anticipated earnings. That way you commit some income to an individual and should be able to follow through and pay them. It might take a few months before you had enough saved to begin doing this. Advertisers love to reach niche audiences, especially audiences with above average incomes. This approach would be mixed with the current approach. If there was money to pay for a review manager it would be offered. If not you would just keep going hoping some generous soul volunteers. -- Brian Wood Ebenezer Enterprises http://webEbenezer.net (651) 251-9384

1 0

Fwd: [Boost-users] [Review] ITL review starts today, February 18th
by Joachim Faulhaber 25 Feb '10

25 Feb '10

Hi, this question of Mathias was posted in both [Boost-users] and [boost] but the answer was accidentally send to [Boost-users] only. ---------- Forwarded message ---------- From: Joachim Faulhaber <afojgo(a)googlemail.com> Date: 2010/2/23 Subject: Re: [Boost-users] [Review] ITL review starts today, February 18th To: boost-users(a)lists.boost.org 2010/2/23 Mathias Gaunard <mathias.gaunard(a)ens-lyon.org>: > Hartmut Kaiser wrote: > >> Interval containers allow to access and manipulate collections of >> intervals >> on an abstract level as sets and maps (of elements), but also to exploit >> segmental information that is given by the interval borders. Interval >> containers are fairly generic. They can be instantiated with discrete >> domain >> (or key) types like integral numeric types or boost::date or time types. >> But >> they also work with continuous domain types like double, rational or >> string. >> Using the latter allows to represent infinite sets and maps. >> >> The interface of the interval containers strives to be very intuitive and >> in >> line with the stl and other boost libraries. >> >> * The set theoretic operations union, difference, intersection >> and symmetric difference are available as operators for all >> interval containers and for many useful overloads between them. > > I have no knowledge of the problem domain (nor did I actually look at the > library yet, sorry about that), but I simply have a question: how does this > compare to geometry libraries or spatial indexes that provide that kind of > facilities for arbitrary dimensions? > > I certainly wouldn't want to force over-generalization to the authors, but I > am wondering how much is gained by limiting things to one dimension. So can this library survive in a field of power called boost of an ever abstracting ever generalizing and ever optimizing bunch of thinkers? A few points: * Geometry usually is based on a numeric coordinate type that comes with a distance measure. Many algorithms in geometry libraries rely on that. Intervals and interval containers of the ITL, similar to SortedAssociativeContainers, only require a strict weak ordering on their key type. So interval containers are less restricted and allow for a larger class of instantiations. You may for example want to work with interval sets of strings. * Interval containers implement interval combining styles that can be useful in providing and manipulating certain time grids. E.g. http://www.herold-faulhaber.de/boost_itl/doc/libs/itl/doc/html/boost_itl/ex… I'm not aware of something like that in geometry. * There are not only interval sets but also interval maps. itl::interval_map is the real workhorse in those real world applications that already exist. interval_maps provide a general idiom of aggregation of associated values. This idiom is very useful for computing statistical data through aggregation that occur in time. * Sometimes a more simple and basic class of data structures is advantageous. It may be simpler to implement optimizations. Like Lukes Manhattan integral polygons compared to more general GGL polygons. * Transfer of geometry knowledge to the more simple models of interval sets and maps may be a step users may not come up with. * The basic set and map concepts of the ITL allow for a variety of applications that may go beyond the scope of geometry. One such example is the self compressing large bitset, which not really has the look and feel of geometry ... http://www.herold-faulhaber.de/boost_itl/doc/libs/itl/doc/html/boost_itl/pr… ... or the example of a system of authorisations. See the example user_groups http://www.herold-faulhaber.de/boost_itl/doc/libs/itl/doc/html/boost_itl/ex… * Well known data structures from geometry like interval_tree or spatial indices do have different focuses and different applications. An interval_tree is more space efficient than an interval_map that can computes all overlaps at once. This may be desired if you need all of them. If you don't and you have reoccurring arbitrary stabbing queries interval_trees are the better choice. * itl::Maps are defining a generalized addition, which I think is a quite powerful feature that accounts for most of the goodies in the applications that we have done. This addition function is a generalisation of the common insert function on SortedAssociativeContainers and it introduces the aggregating properties of itl::Maps. The aggregate on overlap idiom is independent of the question, if the domain_type of the map or interval_map is a scalar or a tuple (has dimensions) or something else. There are already interesting applications of aggregations using itl::maps of tuples, but this exceeds the scope of the core library. Cheers, Joachim

1 0

[date_time] int64 overflow bug with subsecond duration
by Zachary Turner 25 Feb '10

25 Feb '10

Consider the following code, very simple in nature: static const boost::gregorian::date windows_epoch_base(1601, 1, 1); date_time d(windows_epoch_base, microseconds(12908391903991608)); This should resolve to sometime in early 2010 Now, the problem is that this number is already really big, not that many factors of 10 away from the int64 overflow boundary. microseconds() is really just the subsecond_duration<> template, and on my machine it's being parameterized as subsecond_duration<boost::posix_time::time_duration, 1000000>. The code for the constructor is as follows: explicit subsecond_duration(boost::int64_t ss) : base_duration(0,0,0, ss*traits_type::res_adjust() / frac_of_second) {} res_adjust is also resolving to 1000000. So compiler is doing 12908391903991608*res_adjust first, which causes an overflow. I think it should be doing ss * (traits_type::res_adjust() / frac_of_second) instead. I don't know a whole lot about date_time internals, so I thought I would ask here before just submitting a bug. Is the proposed fix correct? I made the change locally and it does indeed appear to resolve the problem. Zach

2 3

[move] will boost::rv work with types with virtual destructor?
by DE 25 Feb '10

25 Feb '10

sorry for a dumb question i noticed boost move lib mentioned in a recent message and decided to look at it i noticed that destructor for boost::rv class is declared but not defined so it can no longer be generated automatically then if an arbitrary type has a virtual destructor and we want to parameterize the boost::rv template with it a vtable for the boost::rv<type> must be generated among other things it must hold a pointer to rv<type>::~rv which is not defined is that really a problem? -- Pavel

2 2