- Boost - lists.boost.org

CVS problems
by Jonathan Turkanis 04 Apr '05

04 Apr '05

Hi, I'm getting "No space left on device" errors, apparently referring to the remote server, while trying to checkout boost via anonymous CVS. Who should I report this to? Jonathan

1 0

CVS problems
by Jonathan Turkanis 04 Apr '05

04 Apr '05

Hi, I'm getting "No space left on device" errors, apparently referring to the remote server, while trying to checkout boost via anonymous CVS. Who should I report this to? Jonathan

1 0

Yet another smart pointer type: a new library proposal
by Marcin Kalici�ski 04 Apr '05

04 Apr '05

Hi Everybody, Smart pointers in boost work on a principle of preventing object destruction while references to it exist anywhere in the system. It causes that whoever owns a boost::shared_ptr to some object, has power to decide for how long will that object live. This is "ownership". In some situations it would be more convenient if object itself decided when to die (owned itself?), and notified all reference holders about this fact, so that they not try to use it any more. A real-life example from a computer game code. There are C++ objects representing tanks on the battlefield. When tank gets hit, it explodes and ceases to exist. The nice, straighforward way to do it is to write "delete this;" in C++. All references to deleted tank which are held by other tanks (for example those currently aiming at it, or following it), will be immediately updated to NULL, provided that they are live_ptrs: template<class T> class live_ptr { // Member functions closely resembling those of shared_ptr // When object pointed to is deleted, this pointer resets its value to NULL }; class Tank; void live_ptr_test() { Tank *tank = new Tank; live_ptr<Tank> p(tank); // not ownership, but a "live" reference delete tank; assert(p == 0); } I have been using (gradually improved) variations of such a smart pointer in some of my projects for last couple of years, and found it useful. Anybody interested? cheers, Marcin Kalicinski

5 6

[test] CodeWarrior+IsDebuggerPresent()
by Stefan Slapeta 04 Apr '05

04 Apr '05

Many tests on CW fail due to a missing IsDebuggerPresent() in execution_monitor.ipp. I assume this code should be disabled for CW, no? http://tinyurl.com/59vnj ### mwcc Compiler: # In: ..\boost\test\impl\execution_monitor.ipp # From: ..\libs\logic\test\tribool_io_test.cpp # ----------------------------------------------- # 199: if( IsDebuggerPresent() ) # Error: ^ # undefined identifier 'IsDebuggerPresent' Stefan

4 10

(Named) Parameter Library
by David Abrahams 04 Apr '05

04 Apr '05

Daniel and I are working on getting the Named Parameter Library up to the standards requested as part of its review. Over the past few months I realized I also needed "unnamed parameters" (which have no definite order and are discriminated by type), and it made sense to extend the library in that direction. There's also a need for unnamed (and maybe even named) template parameters, and all of these facilities can use a lot of code in common with what we've already got. I think it makes sense to just call this library Boost.Parameter and put it in namespace boost::parameters. I thought it made sense to give people a chance to object, just in case. But everyone loves this idea, right? ;-) -- Dave Abrahams Boost Consulting www.boost-consulting.com

4 5

Re: Proposal: intrusive containers
by Jaap Suter 04 Apr '05

04 Apr '05

> I have finished the first version of my STL-like intrusive containers > More information on: > http://people.freenet.de/turtle++/intrusive.html Hi Olaf, I am using intrusive lists a lot in my work, and I would love to see high-quality Boost versions. I have not looked at your code, but I'm impressed with your write-up, and it already addresses many issues. Here are my initial comments and thoughts: 1. Instead of providing the two options of deriving and embedding, let the user provide ADL-style 'next' and 'prev' lookup functions that for a given type provide hooks to the intrusive containers. Look at the boost::intrusive_ptr documentation for more information on this style. You can still provide the existing two methods as possible ways of providing these hooks, but if my legacy code (or for some other reason) has types with their own next and prev pointers, I still want to be able to use your code. 2. As you mention, inserting an object twice leads to problems. In general, intrusive lists need to be handled with more care. I would like to see a diagnostics mode that checks invariants and preconditions. Things to check for would be checking if an instance doesn't already exist on inserts, and checking whether there are no loops in a list. Because this will make list-manipulation really slow, I'd like to be able to enable this on a per-instance basis, probably through boolean-ish or policy-based template parameters. 3. I would love to see a single-linked intrusive container that only supports forward iteration, as well as a list decorator that turns an exiting single or doubly linked list into a ring. Thanks for the work, it looks very promising! Regards, Jaap Suter

3 2

Re: [boost] Re: [hash] Boost.Hash test results not showing at thereports
by JOAQUIN LOPEZ MU?Z 04 Apr '05

04 Apr '05

----- Mensaje original ----- De: Daniel James <daniel(a)calamity.org.uk> Fecha: Lunes, Abril 4, 2005 6:52 pm Asunto: [boost] Re: [hash] Boost.Hash test results not showing at the reports > Joaquín Mª López Muñoz wrote: > > Maybe a "sublibs" file is needed at libs/functional? > > Unfortunately, I can't add that because the functional tests are > not in > a subdirectory. So far now, once the tests are running, they're > going to > appear in 'functional'. It's probably best to live with that until > after > the release. > See my previous reply. The tests are being produced, only they're not shown. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0

Re: [boost] Re: [hash] Boost.Hash test results not showing at thereports
by JOAQUIN LOPEZ MU?Z 04 Apr '05

04 Apr '05

Hi Daniel, ----- Mensaje original ----- De: Daniel James <daniel(a)calamity.org.uk> Fecha: Lunes, Abril 4, 2005 6:50 pm Asunto: [boost] Re: [hash] Boost.Hash test results not showing at the reports > Markus Schöpflin wrote: > > I just discovered the following in the regression run output: > [snip] > > don't know how to make hash_std > > don't know how to make hash_extensions > [snip] > > Maybe this has something to do with the problem? > > That could be it. There's an error in the Jamfile, which for some > reason > doesn't cause any problems when I run the tests locally. > This definitely looks like it should be fixed, but it is not the reason why the tests do not appear at the regression results: some engines are actually running your tests despite that problem, as you can see by manually inspecting: ftp://fx.meta-comm.com/boost-regression/CVS-HEAD/ For instance, Dirxion.zip has - <test-log library="functional" test-name="hash_custom_test" test- type="" test-program="" target- directory="bin/boost/libs/functional/hash/test/hash_custom_test.test/gcc -3_3-darwin/debug" toolset="gcc-3_3-darwin" show-run-output="true"> <link result="succeed" timestamp="2005-04-03 21:22:03 UTC">export DYLD_LIBRARY_PATH c++ -g - o "/Users/stephencarson/Development/boost/results/bin/boost/libs/functio nal/hash/test/hash_custom_test.test/gcc-3_3- darwin/debug/hash_custom_test" "/Users/stephencarson/Development/boost/r esults/bin/boost/libs/functional/hash/test/hash_custom_test.test/gcc- 3_3-darwin/debug/hash_custom_test.o" \ \ - L"/Users/stephencarson/Development/boost/results/bin/boost/libs/test/bui ld/libboost_unit_test_framework.a/gcc-3_3- darwin/debug" "/Users/stephencarson/Development/boost/results/bin/boost/ libs/test/build/libboost_unit_test_framework.a/gcc-3_3- darwin/debug/libboost_unit_test_framework.a" "/Users/stephencarson/Devel opment/boost/results/bin/boost/libs/test/build/libboost_unit_test_framew ork.a/gcc-3_3-darwin/debug/libboost_unit_test_framework.a" \ Chmod1 /Users/stephencarson/Development/boost/results/bin/boost/libs/fun ctional/hash/test/hash_custom_test.test/gcc-3_3- darwin/debug/hash_custom_test chmod 711 "/Users/stephencarson/Development/boost/results/bin/boost/libs/funct ional/hash/test/hash_custom_test.test/gcc-3_3- darwin/debug/hash_custom_test"</link> <run result="succeed" timestamp="2005-04-03 21:22:38 UTC">Running 1 test case... *** No errors detected EXIT STATUS: 0</run> </test-log> So, the problem must lie elsewhere. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0

[optional] Where are the fixes for self-assignment problem?
by Joe Gottman 04 Apr '05

04 Apr '05

Several weeks ago I reported that boost::optional<T> was not safe under self-assignment. The author of the class said he was fixing the library, but no new versions of the code have been checked into the CVS repository. With the code freeze for boost 1.33 only a couple of weeks away, I was wondering whether optional<T> would be fixed on time. Joe Gottman

2 1

[hash] Boost.Hash test results not showing at the reports
by Joaquín Mª López Muñoz 04 Apr '05

04 Apr '05

Seems like the report form is not showing the test results for Boost.Hash, yet some of the engines are already producing them. * The tests are defined at libs/functional/hash/test/JamFile * The tests are the subproject libs/functional/hash/test, and the name of the suite is "functional/hash" * The tests are included in /status/Jamfile with subinclude libs/functional/hash/test ; So everything looks like it is in place, but still the tests don't show. Ideas? Maybe a "sublibs" file is needed at libs/functional? Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

3 3