- Boost - lists.boost.org

[multi_array] Can only disable range checking by defining BOOST_DISABLE_ASSERTS macro
by John Reid 08 Jul '09

08 Jul '09

Defining BOOST_DISABLE_ASSERTS disables boost asserts over the entire compilation. It would be nice to have finer grained control over the range checking in multi_array via a boost.multi_array specific macro.

1 1

[Graph] component_index crash on big graph
by Sandeep Gupta 08 Jul '09

08 Jul '09

Hi, I am using the incremental connected component + component_index to find connected components in an undirected graph. However the program works fine on small small dataset but segfaults on large input. I tried to do some debugging and the best I could come up was that it fails on array_push_front cal at the boost/graph/detail/incremental_component.hpp:78. I have attached the source code and the accompanying small files that contain edgelist. The TrivialEdges.txt works fine but on ManyEdges.txt* the program segfaults. To the best of my understanding the input is sane. Would appreciate any help. * ManyEdges.txt is 320K so please download it from (* http://tinyurl.com/m9blyy) * THis is the dataset I stumbled upon. I apologies for not being able to trim down to smaller size. thanks sandeep

3 7

Universal user interface based only on C++ standard libraries
by Jonathan Enders 07 Jul '09

07 Jul '09

One feature I am always missing when programming in a non-interpreted language like C is the ability to interactively execute the program without writing an entire mechanism for that. This "interpreted when typed"-feature proves itself very useful in python for instance. Although it might be too hard if not impossible to make a "compiled and executed when typed"-feature for C it is possible to just give the user the ability to access a huge amount of the functionality of let's say an object. It's just a lot of work to make all the properties accessible and you might blemish the looks of your actually ui-independent class. A workaround can be applied to lessen the damage. It would consist of just a single function that the class has to implement. That function exports all the functionality (options) to a ui-system which takes care of the command handling and user interaction. This works to the point that it can actually completely replace the ui-design-process (well the ui-system still has to be initialized, 3 lines of code). The work to make the C++ standard library ui-accessible could be taken from the user in advance. As I don't know if any of that is of any relevance to the boost community I will leave it at that. It just seems like a big lack to me that the boost-libraries still have no means of user interaction. If there is any interest in the above I have a working sample which demonstrates what words can only poorly describe. Thanks

8 7

[config] polygon library long long, long double portability
by Simonson, Lucanus J 07 Jul '09

07 Jul '09

I use long long and long double in the configuration of my integer arithmetic traits. I'd like to guard those usages with boost macros to allow the code to compile on platforms that don't support long long and long double (replace with long and double respectively) but I can only find a BOOST_HAS_LONG_LONG macro (which is apparently not documented) and not a corresponding BOOST_HAS_LONG_DOUBLE. Is there some other macro I can use to detect BOOST_HAS_LONG_DOUBLE or a typedef I can use that is already set to long double if available and double otherwise? Thanks, Luke

9 31

[Config] Boost config, long long in xlc 10.1
by Ioannis Papadopoulos 07 Jul '09

07 Jul '09

Hi, 1. I see in config/compiler/vacpp.hpp that versions over 6 are not supported when BOOST_ASSERT_CONFIG is defined: #if (__IBMCPP_ > 600) # if defined(BOOST_ASSERT_CONFIG) # error "Unknown compiler version - please run the configure tests and report the results" # endif #endif Still Boost compiles fine with 10.1, so maybe the number should be changed to 1010, unless there is some other reason that I'm missing. 2. long long is supported through a flag, -qlonglong which also defines _LONG_LONG to be 1. As far as I understand this is not taken into account. Thanks!

2 2

[algorithm] - defect due to accidental ADL
by Neil Groves 07 Jul '09

07 Jul '09

Hello, I have been investigating an issue reported by a user of the RangeEx library proposal. The conclusion was that boost/algorithm/find.hpp suffers from accidental ADL. This was confirmed by protecting against accidental ADL in this file and the user confirming that this solved their problem. Would it be useful to the current maintainer if I went through the source code and fixed similar issues? I would be happy to provide the files for the maintainer to commit. Alternatively is it possible to have agreement and then write access to the repository to fix this problem? Thank you in anticipation, Neil Groves

2 1

[Boost-serialization] Missing file
by K. Noel Belcourt 07 Jul '09

07 Jul '09

Hi, It looks like trunk tests are failing with this error. error: Unable to find file or target named error: 'test_shared_ptr_multi_base.cpp' error: referred from project at error: '../libs/serialization/test' -- Noel

1 0

[pool] Doubling algorithm
by Martin Scheffler 07 Jul '09

07 Jul '09

I just stumbled across an old problem: The maximum size of new allocations cannot be limited in boost::pool. I have a pool that grows to hundreds of megabytes. The pool doubles its allocation size every time it runs out of free space. Because Windows wants to allocate consecutive free space, I get out of memory errors even when my program only uses ~500 mb of main memory. If I could limit the amount of memory to allocate this would not be a problem. This was already discussed here: http://lists.boost.org/Archives/boost/2008/10/143763.php Has anything happened since then? Is boost::pool currently maintained? Or are there alternative memory pool libraries I could use? Thank you! Martin Scheffler

2 1

Ticket Resolutions
by Beman Dawes 07 Jul '09

07 Jul '09

I'd like to add "notabug" as a ticket resolution choice. The use would be for tickets that complain about something that is in fact true, but is by design rather than being a bug. In the past, I've marked these as "invalid". But I dislike doing so because it sounds as if I'm saying there is something wrong with the ticket itself. Comments? --Beman

6 6

Boost.Build question
by Zachary Turner 07 Jul '09

07 Jul '09

I notice the cxxflags for passing command line options to the C++ compiler. However, I need to build the boost libraries themselves in such a way that I can pass command line options to the linker. Is this possible somehow? Thanks

2 1