- Boost - lists.boost.org

[1.35][functional] gcc-4.2:dereferencing type-punned pointer will break strict-aliasing rules
by Mat Marcus 20 Apr '08

20 Apr '08

Hi, In boost 1.35.0 I'm getting many of the following warnings from gcc-4.2.0 under cygwin when using boost/functional/hash: ../boost_libraries/boost/functional/detail/hash_float.hpp: In function 'size_t boost::hash_detail::float_hash_impl(float)': ../boost_libraries/boost/functional/detail/hash_float.hpp:75: warning: dereferencing type-punned pointer will break strict-aliasing rules ../boost_libraries/boost/functional/detail/hash_float.hpp: In function 'size_t boost::hash_detail::float_hash_impl(double)': ../boost_libraries/boost/functional/detail/hash_float.hpp:82: warning: dereferencing type-punned pointer will break strict-aliasing rules ../boost_libraries/boost/functional/detail/hash_float.hpp: In function 'size_t boost::hash_detail::float_hash_impl(long double)': ../boost_libraries/boost/functional/detail/hash_float.hpp:90: warning: dereferencing type-punned pointer will break strict-aliasing rules Is this a boost bug, or a spurious warning? Is there a fix or a sound workaround? Thanks, Mat

7 15

[gil] new io extension
by Christian Henning 20 Apr '08

20 Apr '08

Hi there, there are new tiff image formats readable now. Basically, the new io lib supports up to 4 channels with 1, 2, 4, 6, 8, 10, 12, 14, 16, 24, 32, 64 bits depths. Right now a multi-channel image only supports the same channel depth for all channels. For instance for a rgb10_image_t all channels are 10 bits deep. I would really like to support multi-depths channels but unfortunately there would be too many template instantiations. I would like to ask the boost experts to have a look at my lib and to give some advise on how to deal with it. Also, is there someone that can compile my code on a Linux machine? I don't have access to it and would like to know if all is kosher. Thanks, Christian

1 0

[svn] wave instantiate_cpp_exprgrammer build failure
by Neal Becker 19 Apr '08

19 Apr '08

gcc.compile.c++ bin.v2/libs/wave/build/gcc-4.1.2/release/instantiate_cpp_exprgrammar.o ./boost/wave/util/macro_helpers.hpp: In function ‘boost::wave::token_id boost::wave::util::impl::skip_whitespace(IteratorT&, const IteratorT&)’: ./boost/wave/util/macro_helpers.hpp:262: error: ‘next_token’ was not declared in this scope ./boost/wave/util/macro_helpers.hpp:262: error: expected primary-expression before ‘>’ token ./boost/wave/util/macro_helpers.hpp:262: error: ‘::peek’ has not been declared ./boost/wave/util/macro_helpers.hpp:266: error: expected primary-expression before ‘>’ token ./boost/wave/util/macro_helpers.hpp:266: error: ‘::peek’ has not been declared ./boost/wave/util/macro_helpers.hpp: In function ‘boost::wave::token_id boost::wave::util::impl::skip_whitespace(IteratorT&, const IteratorT&, ContainerT&)’: ./boost/wave/util/macro_helpers.hpp:279: error: ‘next_token’ was not declared in this scope ./boost/wave/util/macro_helpers.hpp:279: error: expected primary-expression before ‘>’ token ./boost/wave/util/macro_helpers.hpp:279: error: ‘::peek’ has not been declared ./boost/wave/util/macro_helpers.hpp:283: error: expected primary-expression before ‘>’ token ./boost/wave/util/macro_helpers.hpp:283: error: ‘::peek’ has not been declared

2 1

[shared_Ptr][tr1] and those dreaded circular dependencies again.
by John Maddock 19 Apr '08

19 Apr '08

A recent change in SVN head (whether directly to shared_ptr or something else I'm not sure) has broken some of the TR1 tests: http://beta.boost.org/development/tests/trunk/developer/output/daw-msvc71-b… The two patches below would fix this, but as ever I'm open to other/better suggestions. Cheers, John. Index: C:/data/boost/boost/trunk/boost/detail/shared_count.hpp =================================================================== --- C:/data/boost/boost/trunk/boost/detail/shared_count.hpp (revision 44538) +++ C:/data/boost/boost/trunk/boost/detail/shared_count.hpp (working copy) @@ -29,7 +29,7 @@ #include <boost/detail/sp_counted_base.hpp> #include <boost/detail/sp_counted_impl.hpp> -#include <memory> // std::auto_ptr +#include <boost/config/no_tr1/memory.hpp> // std::auto_ptr #include <functional> // std::less #include <new> // std::bad_alloc Index: C:/data/boost/boost/trunk/boost/shared_ptr.hpp =================================================================== --- C:/data/boost/boost/trunk/boost/shared_ptr.hpp (revision 44538) +++ C:/data/boost/boost/trunk/boost/shared_ptr.hpp (working copy) @@ -20,7 +20,7 @@ #include <boost/detail/shared_ptr_nmt.hpp> #else -#include <memory> // for std::auto_ptr +#include <boost/config/no_tr1/memory.hpp> // for std::auto_ptr #include <boost/assert.hpp> #include <boost/checked_delete.hpp>

2 2

Some notes on Egg
by John Phillips 19 Apr '08

19 Apr '08

Sorry this is not a full review. I had a half an hour to dedicate to this, and that just isn't enough time for the library. First, I want to say that I find some of the possibilities of this library intriuging, and whatever the outcome of this review I hope to see some form of it in boost in the future. However, I find the documentation difficult to read. Some of this is because of grammatical considerations. For this part of the documentation, I suggest the author works with any of the various people on this list who have offered to assist with ddocumentation in the past. I think there are still some offers with contact information that can be found on the wiki. The other problem I have with the documentation is that the author seems to know what he is talking about well enough that he doesn't realize that it is not complete enough for many potential readers. For this I have two suggestions. The first is to include more examples in the flow of the documentation. Some of them should betrivial, while others show more complete use cases. Every time a new concept or use is introduced, consider how an example could improve the discussion. This has become a hazard in the Boost community because we are all far more steeped in functional programming concepts and practices than the general run of C++ programmers and this leads to a communications disconnect for what we think is clear documentation compared to what the mainstream community needs for clear documentation. My second suggestion is to think a lot about how to reorganize the documentation with a new user of the library in mind. If someone comes to the library without knowing what you are doing, how can you show them each of the advantages of the library, preferably one at a time? How can you build the ddocs to draw them through that process and to make it obvious what parts of the library will help solve what problems for the user? An example is the "How to Read the Docs" section at the bottom of the Quick Start page. The reader is two pages into the docs, and a new reader who doesn't understand will have already given up before getting to this section. Separate it and make it clear where to find it in the ToC, instead of hiding it in another page. I haven't looked at the implementation enough to have a usable opinion, so I can really only talk about the documentation. In my opinion, the docs need substantial improvement before the library should be accepted, and I strongly encourage the author to make those improvements whatever the outcome of the review. John Phillips

2 1

[real_concept] Problem with round() family of functions
by Hartmut Kaiser 19 Apr '08

19 Apr '08

Hi John, while moving the new Spirit to the Boost SVN I came across the new functions introduced to allow rounding and truncation to an integer. Cool, thanks for adding these. But now I have the problem, that these functions are in a different namespace as the original real_concept class, which makes it impossible for the compiler to find these through ADL. Is there any reason why you didn't put the round() and truncate() functions for the real_concept class in the 'namespace concepts' instead? Do I miss something? Regards Hartmut

1 0

Program_options crashing on boost 1.35 on release on MSVC 8.1
by James Talbut 18 Apr '08

18 Apr '08

Hi, The following lines generate an AV on boost 1.35.0 in a release build on MSVC 8 on Vista. boost::program_options::options_description Desc("Allowed Options"); Desc.add_options() ("help,?", "Display help message") ("print_suites", "Print out the suite heirarchy") ("log_level", "all|test_suite|message|warning|error|cpp_exception|system_error|fatal_e rror|nothing") ("result_code", "Set to \"no\" to forces the return code to be zero") ("report_level", "no|confirm|short|detailed") ("run_test", boost::program_options::value< std::string >(), "Test to run") ("save_pattern", "") ("build_info", "Report compiler and build details") ("show_progress", "Makes the framework print progress information.") ("catch_system_errors", "Attempt to catch and report system errors (can make debugging harder)") ("report_format", "HRF|XML") ("log_format", "Set the log format used") ("output_format", "HRF|XML") ("detect_memory_leaks", "Attempt to catch and report memory leaks (can be annoying with 3rd party caches)") ("random", "Run test cases in random order (integer parameter is used as seed)") ("break_exec_path", "") ; It works fine with 1.34 (release or debug) and with a debug build on 1.35. I haven't managed to work out why this is happening yet, so I can't rule out it being my fault. Is anyone aware of anything going on in this area? Jim ________________________________________________________________________ This e-mail, and any attachment, is confidential. If you have received it in error, do not use or disclose the information in any way, notify me immediately, and please delete it from your system. ________________________________________________________________________

2 1

Re: [boost] some bug in serialization/utility.hpp
by Stefan Heinzmann 18 Apr '08

18 Apr '08

Hi wansheg, this is a known bug. See http://svn.boost.org/trac/boost/ticket/1822 for more info. Cheers Stefan __________________________________________________________ Gesendet von Yahoo! Mail. Der Mailbox mit unbegrenztem Speicher. http://de.overview.mail.yahoo.com

1 0

[shifted_ptr] Updated thread and <const T> support
by Phil Bouchard 18 Apr '08

18 Apr '08

Greetings: Upon testing the library on day to day projects, support for <const T> objects is invaluable and thus was added. Support for volatile objects could be important in certain cases also but requires more overloaded functions to be added (const ^2 + volatile ^2). Furthermore I have changed the code to use the new thread_specific_ptr<> object handler but I have disabled it by default with BOOST_SH_DISABLE_THREADS. Thanks, -Phil

2 13

[Regex] A bug with #define BOOST_REGEX_MATCH_EXTRA in using char type
by sevenjay 18 Apr '08

18 Apr '08

Dear all: if you #define BOOST_REGEX_MATCH_EXTRA, using char type will crash, but using string is well. Is that a bug? My environment is Windows XP, C++ Builder 2007. I show the example below: #define BOOST_REGEX_MATCH_EXTRA #include <boost/regex.hpp> #include <iostream> using namespace std; int main() { boost::regex reg("(.*)(\\d{2})"); boost::cmatch m; const char* text = "Note that I'm 31 years old, not 32."; if(boost::regex_search(text,m, reg)) { if (m[1].matched) std::cout << "(.*) matched: " << m[1].str() << '\n'; if (m[2].matched) std::cout << "Found the age: " << m[2] << '\n'; } } It will crash..... If don't #define BOOST_REGEX_MATCH_EXTRA, it works well. And the string type works both well. Maybe this is a bug? thank you for reading.

2 2