LLVM Developers Ponder Using C++11 Features

[phoronix]

Phoronix: LLVM Developers Ponder Using C++11 Features

While C++11 is an ISO standard and the Clang C/C++ compiler front-end to LLVM has been supporting C++11, developers behind the LLVM compiler infrastructure are still deciding whether to allow C++11 language features within their code-base...

http://www.phoronix.com/vr.php?view=MTI3MDM

[kill BSD]

Chris Lattner and all of his LLVM whores should spend their time on better things such as dying or getting bent over and fucked by GCC.

[uid313]

Does the Linux kernel make use of or allow C11?

[Ericg]

Does the Linux kernel make use of or allow C11?

Everything kernel-side is written in pure-C AFAIK. Userspace or shim layers can be C++, but linus and some of the other devs have a real problem with C++

[Ericg]

double post, delete

[uid313]

Everything kernel-side is written in pure-C AFAIK. Userspace or shim layers can be C++, but linus and some of the other devs have a real problem with C++

Yes, I know but I wasn't asking about C++11, I was asking about C11.

[Ericg]

Whoops, sorry, misread. Saw "C" + "11" and my brain automatically dropped in the ++ lol


Linus' thoughts on C11:

Source: http://gcc.gnu.org/ml/gcc/2012-02/msg00042.html

The problem I have with treating this as a C11 issue is that people
aren't using C11 compilers, and won't for a long while.

So quite frankly, it won't be reasonable for the kernel people to say
"use a C11 version" for years to come.

[...]

And don't get me wrong. I don't think that means that C11 is *bad*.
It's just that the kernel is very different from most other projects.
We have to have those crazy architecture-specific header files and
random synchronization macros etc anyway.

C11 is not - I think - even meant to be geared towards the Linux
kernel kind of crazy use. We really do some odd things, adding
compiler features for them is mostly a mistake. asm() takes care of a
lot of the oddities for us, it's not like all of them are about memory
accesses or concurrency either.

I do think that the threading issues in C11 are going to help us
kernel people, because the more people think about issues with
concurrency, they really *will* be hitting some of the issues we've
been having. So it clearly forces clarification of just what the word
"access" implies, for example, which is certainly not going to be bad
for the kernel.