[FoRK] C++11 / STL, Re: Raspberry Pi + Qt5 + OpenGL = awesome

Stephen Williams sdw at lig.net
Fri Apr 12 12:43:21 PDT 2013


On 4/12/13 12:05 PM, Joseph S. Barrera III wrote:
> On 4/12/2013 10:57 AM, Stephen Williams wrote:
> > On 4/12/13 7:57 AM, J. Andrew Rogers wrote:
> >> On Apr 11, 2013, at 9:58 PM, Stephen D. Williams <sdw at lig.net> wrote:
> >>> You should reconsider with modern tools.
> >>>
> >>> I think that is all about the libraries and frameworks, and classes that other people in the project are writing, that are 
> usually the issue.
> >>> Modern C++11, with STL, and/or Qt, is actually nice.
> >>
> >> STL is still broken for some common C++11 tool chains. GCC 4.8, for example.
> >
> > Since STL is effectively part of the language now, C++11 conformance includes STL.  Perhaps you mean that stand lone STL 
> distributions won't work on C++11?  I haven't seen the need so far since it is part of the C++ library now, as far as I've 
> needed. There seems to be a good flow of Boost -> C++ library, although I'm not sure how that will work now that C++11 has 
> adopted modified versions of STL + Boost.
>
> On http://gcc.gnu.org/gcc-4.8/cxx0x_status.html it says "GCC provides experimental support for the 2011 ISO C++ standard." 
> That "experimental" doesn't sound so good.

Why not?  Does this invoke some specific fear?
If your app passes all tests and works, who cares if the compiler is "experimental"?
One group's "experimental" or "beta" is another companies production.  All that matters is regression tests and actual 
experience for the range of needs you have.

>
> http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.2011 shows stuff that's still missing.
>
> Maybe in another year?

Maybe in another year all features will be implemented?  Sure.

Maybe in another year it will be worth taking a look at?  Only if you aren't doing C++, otherwise in most cases switching 
immediately will be far better.
In conciseness, efficiency, and coding speed, C++11 is far better. And you're unlikely to have to rewrite new code since it's 
obvious that everything is going to go to C++11 anyway.

No one implements all C++11 accepted features as far as I know, at least not one of the big three (gcc, clang/llvm, 
VS2012/vc11).  User defined constants for instance are cool but not available yet.  Why wait until all features are there to 
start using the most important features now?

>
> - Joe

sdw



More information about the FoRK mailing list