When could Intel C++ compiler support C++11 feature?

[Zhanghong_T_]

Dear all,

Recently I am trying to compiler some code by Intel C++ Compiler XE 13.0, the code is somewhat like the follows:

            for(uint i=0; i<skeleton.size(); i++) {
                for(auto &entry : skeleton) {
                    action(i, entry.vid, entry);
                }
            }

when compiling by Intel C++ Compiler XE 13.0, the following error displayed:

Error    17    error : cannot deduce "auto" type (initializer required)  

I got to know that this new feature is not supporttd by latest Intel C++ compiler. When could Intel C++ compiler support this feature?

Thanks,

Zhanghong Tang

[Marián__VooDooMan__M]

Greetings,

As Intel's support person said recently, C++11 is the main goal, though, (IIRC it was Jeniffer) said thay cannot comment details for future Intel's plans.

PS: I am waiting impatiently for new release, since in recent one /MP switch under Windows is broken, so I'd like to have knowledge of the timeline, but as she said, they cannot provide comments about this...

[SergeyKostrov]

Did you try to compile the test-case with option /Qstd=c++0x? Also, Please follow a link to see all supported C++11 features: software.intel.com/en-us/articles/c0x-features-supported-by-intel-c-compiler

[Zhanghong_T_]

Dear Dr. Sergey Kostrov,

Thank you very much for your instructioin, this option works for my issue.

Thanks,

Zhanghong Tang

[Judith_W_Intel]

Yes the "auto" keyword and c++11 range based for loops (which is what this code is using) should be supported in C++ 13.0 if you use the option to turn on c++11 (-std=c++11 on Linux and /Qstd=c++11 on Windows).

Judy

[Zhanghong_T_]

Dear Dr. Sergey Kostrov and Judith Ward,

Thank you very much for your kindly reply. I tested and found that both two options /Qstd=c++0x and /Qstd=c++11 are OK for me. Could you please tell me what's the difference between these two options?

Thanks,

Zhanghong Tang

[TimP]

The effect of those options should be identical.  c++0x was chosen by gcc for early development; after the standard was confirmed, c++11 was recommended as the spelling for the future.

[Zhanghong_T_]

Dear Dr. Timp,

Thanks again for your kindly explaination.

I suggest that this option be turned on by default.

Thanks,

Zhanghong Tang

[SergeyKostrov]

>>...I suggest that this option be turned on by default. Thanks for your suggestion. I think this is a very dangerous thing and a developer should decide if the option needs to be set on. In general, command line options of any C++ compiler are "switchers" and that approach is very flexible. If some default option is changed it could bring lots of unpredictable issues and problems with existing hundreds of millions C/C++ code lines already implemented, tested, verified, validated, stabilized, frozen, etc.

[Judith_W_Intel]

Turning it on by default would break compatibility with the Microsoft compiler. We only enable those c++11 features by default which are already implemented by the reference Microsoft compiler on the user's system.