Summary of C++0x Feature Availability
in [C++]
|
Prev: Is "<unsigned type here> var = -1;" specified behavior?
Next: I keep running into long term c++ programmers who refuse to use exceptions
From: Scott Meyers on 16 Mar 2010 14:58 Herb Sutter wrote: > I'm just glad that Microsoft isn't at the very back of the conformance > pack this time. :-) Since I'm on record for taking MS to task for its lousy conformance last time 'round ( http://www.artima.com/cppsource/top_cpp_software.html ), I'm happy to say that I've been very pleased with the C++0x features (both language and library) available in the freely-available VC10. (My primary experience is with the betas, but the RC is out now at http://msdn.microsoft.com/en-us/vstudio/dd582936.aspx .) The combination of auto, lambdas, rvalue references, nullptr, and nearly the entire set of library extensions has been a true pleasure to work with. It'd be sweet to have the concurrency stuff, too, but nobody has that yet, as far as I know. Between gcc 4.4 and VC10 (the two compilers I've been experimenting with), I don't think there is a reasonable way to say which one is "more" conformant. Of the two, only MS has lambdas, for example, while only gcc has variadic templates, default and deleted functions, and uniform initialization. For playing around purposes (i.e., where you try lots of different things and combinations of things to see how they work), I prefer VC10. My preference may change once gcc 4.5 comes out, but right now, my subjective opinion is that far from trailing the pack, MS is arguably leading it. To have such an influential vendor putting significant effort into implementing the nascent revised standard is a boon to all people interested in C++ development. From a standards-conformance point of view, I'd say that VC10 makes up for VC6. (Which I guess means that from my perspective, 10 isn't the new 6, it's the new anti-6.) Scott -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: johannes.dahlstrom on 18 Mar 2010 07:52 On Mar 17, 7:58 am, Scott Meyers <NeverR...(a)aristeia.com> wrote: > It'd be sweet to have the concurrency stuff, too, > but nobody has that yet, as far as I know. g++ 4.4 has threads, atomics, locks, mutexes, and condition variables. Alas, not yet any of the really fancy stuff like futures. Please take a look at the example I hobbled together a while ago: http://lilja.asteriski.fi/~sharlin/procon3.cpp Johannes -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Scott Meyers on 18 Mar 2010 18:11 johannes.dahlstrom(a)gmail.com wrote: > g++ 4.4 has threads, atomics, locks, mutexes, and condition variables. Hmmm. Given #include <thread> #include <mutex> int main() { std::thread t; std::mutex m; } I get: g++ (TDM-1 mingw32) 4.4.1 Copyright (C) 2009 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. threads.cpp: In function 'int main()': threads.cpp:6: error: 'thread' is not a member of 'std' threads.cpp:6: error: expected ';' before 't' threads.cpp:7: error: 'mutex' is not a member of 'std' threads.cpp:7: error: expected ';' before 'm' > Alas, not yet any of the really fancy stuff like futures. Please take > a look at the example I hobbled together a while ago: > > http://lilja.asteriski.fi/~sharlin/procon3.cpp My copy of gcc can't compile your code, either: g++ (TDM-1 mingw32) 4.4.1 Copyright (C) 2009 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. jd.cpp:28: error: expected type-specifier jd.cpp:28: error: expected '>' jd.cpp:35: error: ISO C++ forbids declaration of 'unique_lock' with no type jd.cpp:35: error: typedef name may not be a nested-name-specifier jd.cpp:35: error: expected ';' before '<' token jd.cpp:47: error: type 'work_pool<T, Container, Mutex>' is not derived from type 'work_pool<T, Container, Mutex>::handle' jd.cpp:47: error: expected ';' before 'lock_type' jd.cpp:126: error: 'lock_type' does not name a type jd.cpp:137: error: 'condition_variable' in namespace 'std' does not name a type jd.cpp: In constructor 'work_pool<T, Container, Mutex>::handle::handle(work_pool<T, Container, Mutex>::handle&&)': jd.cpp:51: error: class 'work_pool<T, Container, Mutex>::handle' does not have any field named 'lock' jd.cpp: In constructor 'work_pool<T, Container, Mutex>::handle::handle(work_pool<T, Container, Mutex>&)': jd.cpp:123: error: class 'work_pool<T, Container, Mutex>::handle' does not have any field named 'lock' jd.cpp: At global scope: jd.cpp:145: error: template argument 3 is invalid jd.cpp: In function 'void produce(int, int&, const std::atomic<bool>&)': jd.cpp:152: error: 'std::this_thread' has not been declared jd.cpp:155: error: request for member 'lock' in 'pool', which is of non-class type 'int' jd.cpp:155: error: unable to deduce 'auto' from '<expression error>' jd.cpp: At global scope: jd.cpp:169: error: template argument 3 is invalid jd.cpp: In function 'void consume(int, int&, const std::atomic<bool>&)': jd.cpp:176: error: request for member 'lock' in 'pool', which is of non-class type 'int' jd.cpp:176: error: unable to deduce 'auto' from '<expression error>' jd.cpp:187: error: 'std::this_thread' has not been declared jd.cpp: In function 'int main()': jd.cpp:196: error: template argument 3 is invalid jd.cpp:196: error: invalid type in declaration before '(' token jd.cpp:200: error: 'thread' was not declared in this scope jd.cpp:200: error: template argument 1 is invalid jd.cpp:200: error: template argument 2 is invalid jd.cpp:200: error: invalid type in declaration before ',' token jd.cpp:204: error: request for member 'emplace_back' in 'consumers', which is of non-class type 'int' jd.cpp:204: error: 'cont' was not declared in this scope jd.cpp:205: error: request for member 'emplace_back' in 'producers', which is of non-class type 'int' jd.cpp:208: error: 'this_thread' has not been declared jd.cpp:214: error: request for member 'begin' in 'producers', which is of non-class type 'int' jd.cpp:214: error: request for member 'end' in 'producers', which is of non-class type 'int' jd.cpp:214: error: 'thread' is not a class, namespace, or enumeration jd.cpp:215: error: request for member 'begin' in 'consumers', which is of non-class type 'int' jd.cpp:215: error: request for member 'end' in 'consumers', which is of non-class type 'int' jd.cpp:215: error: 'thread' is not a class, namespace, or enumeration Any idea why our results are different? Scott -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Paul Bibbings on 19 Mar 2010 16:18 Scott Meyers <NeverRead(a)aristeia.com> writes: > johannes.dahlstrom(a)gmail.com wrote: >> g++ 4.4 has threads, atomics, locks, mutexes, and condition variables. > > Hmmm. Given > > #include <thread> > #include <mutex> > > int main() > { > std::thread t; > std::mutex m; > } > > I get: > > g++ (TDM-1 mingw32) 4.4.1 > Copyright (C) 2009 Free Software Foundation, Inc. > This is free software; see the source for copying conditions. There is NO > warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. > > threads.cpp: In function 'int main()': > threads.cpp:6: error: 'thread' is not a member of 'std' > threads.cpp:6: error: expected ';' before 't' > threads.cpp:7: error: 'mutex' is not a member of 'std' > threads.cpp:7: error: expected ';' before 'm' <snip /> > Any idea why our results are different? > > Scott I see you're attempting to compile under MinGW. Try: #include <thread> #include <iostream> int main() { #if !defined(_GLIBCXX_HAS_GTHREADS) std::cout << "glib doesn't have gthreads on my platform\n"; #endif } In the gcc c++0x headers, _GLIBCXX_HAS_GTHREADS is a wrapper for the new threading capabilities. I don't know the exact details, but I have the same problems with gcc-4.4.3 (i686-pc-cygwin-gcc) on Windows. Obviously implementations of gcc require glib to have gthreads support, but I have yet to find (or build) an implementation for Windows - whether under Cygwin or MinGW - that can achieve this. As for the precise reasons why gthreads support seems not to be available under Windows (if that's not too bold an assertion), I have not yet gone that far in my investigations. Regards Paul Bibbings -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: johannes.dahlstrom on 19 Mar 2010 16:17
On Mar 19, 11:11 am, Scott Meyers <NeverR...(a)aristeia.com> wrote: > > Any idea why our results are different? > My compilation environment: $ uname -a; gcc -v Linux aurora 2.6.31-20-generic-pae #58-Ubuntu SMP Fri Mar 12 06:25:51 UTC 2010 i686 GNU/Linux [snip] Thread model: posix gcc version 4.4.1 (Ubuntu 4.4.1-4ubuntu9) It seems that MinGW, being "minimal", doesn't have a pthreads emulation layer and thus none of the library components dependent on pthreads, either (and nobody has yet written a libstdc++0x implementation using native Windows threads...) The Cygwin project at http://cygwin.com/ offers a much more complete POSIX environment including pthreads, but they don't have GCC 4.4 binaries available yet. Oh well. The GCC folks claim that vanilla GCC compiles cleanly on Cygwin should you like to get adventurous... Johannes -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] |