- Boost - lists.boost.org

[task] functor passed to a boost::task::task not always destroyed?
by Harald Winroth 25 Aug '09

25 Aug '09

I wrote a short test program to check that each copy construction of a functor passed to a boost::task::task and boost::task::async was balanced by a corresponding destructor call. This is indeed the case when the task is run in a thread pool but it seems as if one destructor call is missing when the task is submitted to to boost::task::async with a boost::task::new_thread execution policy. Here is the a short test program and the resulting output. For comparison, the same functor is also passed directly to a boost::thread, which behaves as expected. The program was compiled using Visual C++ 2005 (SP1), and boost 1.39.0 with the latest Boost.Task library from the vault dropped in. Any ideas? Am I missing something? Best regards Harald --- Output: Running Runnable as a task Runnable() [0012FF57] Runnable(const Runnable&) [0012FE88] Runnable(const Runnable&) [0012FE40] Runnable(const Runnable&) [00463B9C] ~Runnable() [0012FE40] ~Runnable() [0012FE88] Runnable::operator() [00463B9C] ~Runnable() [0012FF57] Running Runnable directly on a thread Runnable() [0012FF33] Runnable(const Runnable&) [0012FE8C] Runnable(const Runnable&) [0012FE5C] Runnable(const Runnable&) [0012FDDC] Runnable(const Runnable&) [001541D8] ~Runnable() [0012FDDC] ~Runnable() [0012FE5C] ~Runnable() [0012FE8C] Runnable::operator() [001541D8] ~Runnable() [001541D8] ~Runnable() [0012FF33] --- Test program: #include <boost/thread.hpp> #include <boost/task.hpp> #include <iostream> void print(const std::string& s, void* instance = 0) { static boost::mutex mutex; boost::lock_guard<boost::mutex> guard(mutex); std::cout << s; if (instance) std::cout << " [" << instance << "]"; std::cout << std::endl; } struct Runnable { Runnable() { print("Runnable()", this); } Runnable(const Runnable&) { print("Runnable(const Runnable&)", this); } ~Runnable() { print("~Runnable()", this); } void operator()() { print("Runnable::operator()", this); } }; int main() { { print("Running Runnable as a task"); Runnable r; boost::task::handle<void> h = boost::task::async(boost::task::make_task(r), boost::task::new_thread()); h.wait(); } { print("Running Runnable directly on a thread"); Runnable r; boost::thread thread(r); thread.join(); } return 0; }

2 1

[1.40.0] Freeze for release candidate
by Beman Dawes 25 Aug '09

25 Aug '09

As far as I can tell, all the branches/release merges we were waiting for have now been completed. The plan is to go ahead with the release candidate as soon as all the release branch tests have cycled for revision 55757. Please commit no more changes to branches/release without explicit permission from me. Thanks, --Beman

9 32

Re: [boost] [1.40.0] Freeze for release candidate
by Eric Niebler 25 Aug '09

25 Aug '09

Matthias Troyer wrote: > On Aug 25, 2009, at 3:15 PM, Eric Niebler wrote: >> We have some MPI test results now. Matthias >> should have a look and report on the failures there. Then we can >> decide whether there are any showstoppers. >> >> Marshal, can you make sure Matthias has access to your test results? > > I saw some test results, and they are the failures I also sometimes see: > the skeleton&contents tests sometimes fail, and sometimes work. This is > the same as in the past and Doug Gregor had explained it as a problem of > the underlying MPI implementation. Looking deeper into it would require > more time than is available. These are not regressions but issues that > have been around for a while. Thanks, Matthias. This is good enough for me to feel comfortable moving ahead with the 1.40 release. Beman, Jeremiah, any objections? -- Eric Niebler BoostPro Computing http://www.boostpro.com

3 4

[move][containers] Fast track request response
by John Phillips 25 Aug '09

25 Aug '09

Ion and all, Ron and I have discussed the fast track review requests, and we have decided they should be normal reviews. The Move library is a single header, but that header has a lot of functionality in it. In addition, it is a very good capability to provide so we expect that it will be used heavily from the time it is added until move support becomes commonly implemented in C++0x compliant compilers. So, giving everyone a chance to really dig deep and mull the details is important. The container library is a larger collection of code and will just take some time for people to absorb well enough to review well and so should have a full review. To be clear, this is not intended as a criticism of the quality of the code and documentation, but instead a recognition of how important it is likely to become. So, we want to make sure it gets a thorough review. Thanks to Ion for producing the libraries, and to the future reviewers for their input. We strongly encourage review manager volunteers to step forward so we can schedule the reviews quickly. John Phillips Ron Garcia Review Wizards

2 1

[UUID] uuid.hpp Objective C++ conflict.
by M M 25 Aug '09

25 Aug '09

Hello, I'm interested in using the uuid library for iphone development. Unfortunately, I can't use it "out of the box" because the function nil in uuid.hpp collides with the Objective C keyword of the same name and, hence, I get a compiler error. If I change the name to mynil, then it compiles ok. 1) Given this conflict, will you still proceed with the same function name? (As an fyi, all other boost libraries I've incorporated into the iphone fortunately don't have this problem.) 2) If so, do you have any suggestions on how I can handle this issue besides manually changing the function name from nil to something else? Thanks, Mostafa

6 9

[UUID] uuid.hpp ODR violation and Ctrl-M pollution
by Mostafa 25 Aug '09

25 Aug '09

Hi all, Using "uuid_alt_v13_pod.zip": 1) It seems that some namespace function definitions in uuid.hpp violate the ODR, specifically the ones lacking inline keyword. This is one of the errors I get using Apple's g++-4.0 when including uuid.hpp in two different source files: ld: duplicate symbol boost::uuids::operator>=(boost::uuids::uuid const&, boost::uuids::uuid const&) If I add inline to the aforementioned function definitions, then the linker errors go away, which makes since. After some digging around, this issue was discussed beginning with: http://lists.boost.org/Archives/boost/2007/05/122479.php but seems to have cropped back up. Examples are: "uuid/boost/uuid/uuid_io.hpp" and the files under "uuid/libs/uuid/test/"2) Also some files in the package are pseudo-randomly sprinkled with Ctrl-M characters, as if someone edited them with two different editors.. -Mostafa

3 4

[1.40] Beta build of PDF docs available.
by John Maddock 25 Aug '09

25 Aug '09

A beta build of the docs for Boost-1.40 in PDF format is available from https://sourceforge.net/projects/boost/files/boost-docs/1.40.0/boost_pdf_1_… If folks can check their library's PDF docs for the usual SNAFU's that would be much appreciated (note only if your Boost library contains quickbook/boostbook/docbook format docs). Enjoy, John.

5 13

Re: [boost] Boost.Multimethod proposal
by OvermindDL1 25 Aug '09

25 Aug '09

On Sat, Aug 22, 2009 at 10:40 PM, Kim Barrett<kab.conundrums(a)verizon.net> wrote: > At 11:58 PM -0400 8/22/09, Edward Diener wrote: >> >> This does not explain the practical purpose of multimethods. I know what >> virtual functions are, of course. What in multimethods improves on the >> polymorphic capabilities of virtual functions that make them a practical >> choice for use over normal polymorphism. > > The previously referenced paper > (http://research.att.com/~bs/multimethods.pdf) > contains rationale discussion, including several well chosen examples where > multiple dispatch can be beneficial. To be blunt though, Virtual Methods are not proper OO, multimethods are proper OO. C++ has been lacking in this, sometimes important area of OO, where things like LISP do wonderfully. Hmm, I know LISP has a well optimized multimethod implementation written in pure LISP, wonder how easily that method could be transferred to C++, although more verbose...

4 6

Request for feedback - Map providing insertion order preserving iteration
by Gautam Sewani 24 Aug '09

24 Aug '09

Hi, We wanted to add a map to Boost C++ libraries which would allow iteration in the order in which elements were inserted, and would also provide logarithmic time retrieval. A simple design for this would be to have a balanced binary tree along with an auxiliary linked list of pointers to the keys in the tree. We would be grateful if you could provide feedback on the utility and design. Regards Gautam Sewani Satyam Shekhar

2 1

[signals2] How to build docs?
by John Maddock 24 Aug '09

24 Aug '09

I'm looking into building PDF versions of the docs for 1.40 and I'm having trouble building the signals2 docs. For example if I cd into libs/signals2/doc and build from there, then the XML code snippets aren't found. Likewise if I build as a dependency from doc/pdf. Any ideas? Thanks, John.

3 7