LLVM 3.1 Has Been Quietly Postponed

[phoronix]

Phoronix: LLVM 3.1 Has Been Quietly Postponed

The major v3.1 update to the LLVM and Clang compiler components were quietly delayed last week. There's still no official communication on this setback for the Apple-sponsored compiler technology...

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

[ChrisXY]

http://lists.cs.uiuc.edu/pipermail/l...ay/049895.html

[hwertz]

That mailing list post is exceptionally unhelpful (not directed at ChrisXY, it *is* helpful for you to post the official response...) I would assume the real situation is there is a showstopper bug; giving a hard date then would be pointless since it'd simply ship when the bug is fixed. But it is Apple so who knows...?

[blinxwang]

Maybe they finally realized LLVM is a pointless endeavor and are trying to find a way to explain to the community that it's going to be dead.

[septianix]

Maybe they finally realized LLVM is a pointless endeavor and are trying to find a way to explain to the community that it's going to be dead.

Sorry to crush your hope!

http://lists.cs.uiuc.edu/pipermail/l...ay/000041.html

[ChrisXY]

I would assume the real situation is there is a showstopper bug; giving a hard date then would be pointless since it'd simply ship when the bug is fixed. But it is Apple so who knows...?

There has been a "final" svn tag for several days. I would assume that it was ready all along and they realized "maybe somebody should write some release notes and assemble all the changes we have made since 3.0"... and nobody did.

[hwertz]

Maybe they finally realized LLVM is a pointless endeavor and are trying to find a way to explain to the community that it's going to be dead.

But it's not pointless. It's used within mesa for shader programs and such, it's used for GPGPU purposes (compiling OpenCL or C or whatever into code to run on the GPU).

Since both plain gcc, and LLVM, are really more a framework where various optimizations are developed, they should be able to feed off each other's successful strategies; as long as GCC and LLVM don't go stagnant, I really expect them to leapfrog each other in terms of speed. I saw a post a while back where someone reported LLVM 2.8 was faster than the earlier gcc 4.x version they were testing against, now LLVM 3.1 is slower than gcc-4.6.