The State Of GCC 4.7.0: Still Months Away

[phoronix]

Phoronix: The State Of GCC 4.7.0: Still Months Away

Red Hat's Jakub Jelinek has issued a new status update concerning the state of the GCC 4.7 compiler...

http://www.phoronix.com/vr.php?view=OTkwMg

[cl333r]

Hopefully solid support for C++11 won't arrive by like 2017.

What's your guess folks when this might happen?

[JakubJelinek]

Hopefully solid support for C++11 won't arrive by like 2017.

What's your guess folks when this might happen?

See http://gcc.gnu.org/gcc-4.7/cxx0x_status.html for the current state. There is progress being done about the C++0x memory model and also e.g. user defined literals.

[snogglethorpe]

Hopefully solid support for C++11 won't arrive by like 2017.

In general I think it's not so useful to think of C++11 support as a single lump item—few compilers will have complete or near complete C++11 support any time soon.

Instead, think of "these features I wanna use, X, Y, and Z" and start using them whenever the compilers you use have support for them.

gcc already (even in older gcc releases) supports many of the most useful C++11 features, e.g., constexpr, rvalue refs, lambda, initializer lists, noexcept, auto, range-based-for, threads, much of the library (I love the random stuff!). The atomic/memory-model stuff is ongoing, but it seems like there will be useful parts of it in the near future.

The main things I want but which aren't supported yet are the memory-model stuff (seems like it's progressing well) and delegating/inheriting constructors (not sure about this).

[ciplogic]

Sorry to not know such basic things, but what means LRA and LRA in GCC?

Vladimir Makarov also responded to say that the GCC LRA support is a long project and that it won't be ready for GCC 4.7

[snogglethorpe]

Sorry to not know such basic things, but what means LRA and LRA in GCC?

"Local Register Allocation"

http://vmakarov.fedorapeople.org/LRA.html

(I don't know so much about the details, but my impression is that it's intended to replace the "reload" phase of gcc, which is wonderful, because I do know very well that reload is one of the most horrible unmaintainable insane and problematic parts of gcc...)

[ciplogic]

Thank you! Really amazingly clear link and description!