- Boost - lists.boost.org

Regressions: timeout for compile and run
by Stefan Slapeta 28 Jul '04

28 Jul '04

CW 9 and Intel 8 regressions did not run (better: finish) for some days because I've been away and had no chance to confirm the popups of assertion failures which were produced by some test cases!@!@! Is there no possibility to set a timeout after the compile/run process is killed by bjam? I don't see any sense behind giving a test case more than 5 minutes to run! Stefan

4 15

[range] doc nit
by Jonathan Turkanis 28 Jul '04

28 Jul '04

Thorsten, In a couple of places the docs for Boost.Range talk about 'nested typedefs': 1) In the first paragraph of boots_range.html: "The subset deals mostly with iterator returning functions and nested typedefs." 2) In the first paragraph of external_concepts.html: "The concept can be specified as a set of member functions with well-defined semantics and a set of nested typedefs with well-defined properties." In both cases, I think it would be more appropriate to talk about 'nested types.' The subject of the first sentence is standard library containers, whose iterators can be nested classes. The second sentence is about generic programming in general so the same considerations should apply. The article 'Generic Programming Techniques' says that associated types are 'typically' member typedefs or associated externally throught traits, which I think is slightly more accurate. Jonathan

2 1

Off topic:making replies threaded
by Tonko Juricic 28 Jul '04

28 Jul '04

I use http://news.gmane.org/gmane.comp.lib.boost.devel to read group archives and I send my replies from Yahoo mail. After trying several variations on mail Subject prefixes I still can not make my replies appear as the part of the thread, when viewed from Gmane. I am sure some will find those 'detached' replies as annoying as I do so any suggestion on how to fix it would be much appreciated. Tony __________________________________ Do you Yahoo!? Yahoo! Mail - You care about security. So do we. http://promotions.yahoo.com/new_mail

3 2

filesystem thoughts/issues
by Jeff Flinn 28 Jul '04

28 Jul '04

I'm now getting into using the filesystem library. Of course the first thing I've run into is the need to generate lexically relative paths. Having the luxury of being able to discount links, I have a working solution, atleast for windows. Some issues that arose are related to a path as a sequence container. While not show stoppers, the lack of some sequence container member functions - such as size(), assign(itr,itr) - required that existing templated classes be refactored to allow use of path as a template argument. Additionally lacking is a const_iterator type. size() is used to determine the 'shorter' of two paths whose path::iterator are passed to std::mismatch as the controlling (beg,end] arguments. The pair of iterators returned by std::mismatch are used to populate the common, first_remainder and second_remainder paths. If there is interest in this code I can post it. Any thoughts on providing additional sequence container member functions and typedefs in a post 1.32 release? Also any plans on incorporating Richard Johnson's globiter facilities? It would be nice to have a filtered_directory_iterator in any case. Something like: filtered_directory_iterator ( const fs::path& , boost::function1<bool,const std::string&> ) as a convenience class based on the boost::filter_iterator. Then I could: std::for_each ( filtered_directory_iterator( aPath, bind( extension, _1 ) == ".hpp" ) , filtered_directory_iterator() , SomeFnc ); Then perhaps 'globbing' would be better supported as an additional regex syntax_option_type? This would keep filesystem from depending on regex. Jeff F

1 0

Range/compat2 is not correct
by Stefan Slapeta 28 Jul '04

28 Jul '04

compat2.cpp is failing on VC 7.1 and the compiler seems to be right: there is a 'typename' missing! Stefan ..\libs\range\test\compat2.cpp(45) : warning C4346: 'range_iterator<Container>::pts<C>::type' : dependent name is not a type prefix with 'typename' to indicate a type

2 2

difference_type for int in numeric_traits
by DY, JERRY U (SBCSI) 28 Jul '04

28 Jul '04

In boost/detail/numeric_traits.hpp, integer_traits can be used to find the difference_type for an int type, in aC++, I get a long long -- I don't think that's the correct result, but can somebody please tell me why not? Thanks! integer_traits<int>::difference_type // long long?? I expect this to be an int Here's why: template <class Integer, bool = true> struct integer_traits { ... typedef typename if_true<(int(x::is_signed) // true && (!int(x::is_bounded) // true || (int(x::digits) + 1 >= digit_traits<boost::intmax_t>::digits))) // false >::template then< Integer, typename if_true<(int(x::digits) + 1 < digit_traits<signed int>::digits)>::template then< // false signed int, typename if_true<(int(x::digits) + 1 < digit_traits<signed long>::digits)>::template then< // false signed long, // else intmax_t // this is long long >::type>::type>::type difference_type; ... } -Jerry

1 0

Missing '\' in boost/serialization/export.hpp ?
by Jonathan Wakely 28 Jul '04

28 Jul '04

Should there be a backslash at the end of line 235 in boost/serialization/export.hpp ? jon

1 0

Intel: uBLAS and range go DR, random still unresolved
by Stefan Slapeta 28 Jul '04

28 Jul '04

I've submitted a Intel 8 DR for two obvious compiler defects: test2 in uBLAS and the iterator_range in iterator (ICE). Is there nobody feeling competent/responsible to give any hint for the random failure? I don't know if I can find it on my own. Stefan

1 2

VC 7.1 warning in linear_congruential.hpp
by Stefan Slapeta 28 Jul '04

28 Jul '04

in rand48::operator () [...]\linear_congruential.hpp(218) : warning C4244: 'return' : conversion from 'boost::uint64_t' to 'boost::int32_t', possible loss of data Patch attached. Stefan RCS file: /cvsroot/boost/boost/boost/random/linear_congruential.hpp,v retrieving revision 1.18 diff -u -r1.18 linear_congruential.hpp --- linear_congruential.hpp 27 Jul 2004 03:43:32 -0000 1.18 +++ linear_congruential.hpp 28 Jul 2004 15:09:10 -0000 @@ -215,7 +215,7 @@ void seed(uint64_t x0) { lcf.seed(x0); } template<class It> void seed(It& first, It last) { lcf.seed(first,last); } - int32_t operator()() { return lcf() >> 17; } + int32_t operator()() { return static_cast<int32_t>(lcf() >> 17); } // by experiment from lrand48() static bool validation(int32_t x) { return x == 1993516219; }

1 0

RE: [boost] Boost Test Library: Bug in floating-point comparison?
by Rozental, Gennadiy 28 Jul '04

28 Jul '04

> Hi, > > I want to use the boost::test floating-point comparison > algorithms (Boost > 1.31.0) and wrote the following simple lines of code: > > double d1 = 0.2; > double d2 = 0.8; > double d3 = d2-d1-d1-d1-d1; // d3 should be zero > > bool b = boost::test_toolbox::check_is_closed( d3, 0., 1e-3 ); > > To my surprise, the return value b is FALSE, although d3 > should be _very_ > close to zero (and is 5.55112e-17 on my machine). > > Am I missing something or is this a bug in the test library? > > Thanks, > Michael > > -- > Michael Kettner > kettner(a)ive.uni-hannover.de Read algorithm description. The algorithm compare "relative errors". In your case it's 1 and infinity. Definitely more then 1e-03. It you want to compare absolute values (which is not recommended in real life applications) you either could write you own predicate or wait a bit, while I am considering adding BOOST_IS_SMALL tool that will be responsible namely for this kind of test. Gennadiy.

2 1