FreeBSD Jumps Quickly On LLVM/Clang 3.2

[phoronix]

Phoronix: FreeBSD Jumps Quickly On LLVM/Clang 3.2

While just released on Friday, FreeBSD has already pulled LLVM/Clang 3.2 into its "head" repository and will be pushing it into the FreeBSD 9/Stable series in the weeks ahead...

http://www.phoronix.com/vr.php?view=MTI2MDU

[BO$$]

Why? Isn't clang shittier than gcc?? These bsd guys really enjoy having worse performance than linux.

[deanjo]

Why? Isn't clang shittier than gcc?? These bsd guys really enjoy having worse performance than linux.

The small performance difference between GCC and Clang is made up for in other areas such as debugging.

[Ericg]

Why? Isn't clang shittier than gcc?? These bsd guys really enjoy having worse performance than linux.

"Shittier" is a pretty vague and worthless term for a compiler. The phrase youre looking for is "Doesnt clang produce slower binaries than gcc?" In which case....mainline clang vs mainline gcc? Sure, yes it does. But *BSD has been stuck at mainline clang vs gcc 4.3 I think, and at THAT level I'm pretty sure clang matches or beats gcc.

Personally I'm looking forward to LLVM just because its a new, cleaner, code base, better debugging capacities (I know gcc is working on getting better but we'll see), more modular (see above), and integrates a lot better into IDE's because of the modularity of it.

If Apple wants to take mainline LLVM and add on some optimizations for closed source...fine, thats their right via the license. But I really don't think that Apple will take mainline LLVM and have some drastic differences vs mainline and their version it-- it adds more work for them to maintain those diff's and to make sure they stay current, usable and valid.

[smitty3268]

The small performance difference between GCC and Clang is made up for in other areas such as debugging.

It's obvious the 1st, 2nd, and 3rd reasons they are using Clang is because of the BSD license.

And there's nothing wrong with that. Remember all the fuss GPL guys put up about Linus using a proprietary code repository software rather than svn/cvs? That turned out pretty well in the long run.

[ryao]

It's obvious the 1st, 2nd, and 3rd reasons they are using Clang is because of the BSD license.

And there's nothing wrong with that. Remember all the fuss GPL guys put up about Linus using a proprietary code repository software rather than svn/cvs? That turned out pretty well in the long run.

Clang is not under the BSD license. It is under the UoI-NCSA license, which is BSD-style, but different:

http://sources.gentoo.org/cgi-bin/vi...SA?view=markup

[deanjo]

It's obvious the 1st, 2nd, and 3rd reasons they are using Clang is because of the BSD license.

That maybe the case for inclusion into the distro but that is not the primary reasons for the actual end users of Clang who just want a good modern compiler. Users of FreeBSD are less concerned about licensing issues then they are of the technology.

[XorEaxEax]

That maybe the case for inclusion into the distro but that is not the primary reasons for the actual end users of Clang who just want a good modern compiler. Users of FreeBSD are less concerned about licensing issues then they are of the technology.

Users may or may not be, but the FreeBSD foundation certainly isn't putting technology above licencing. That is why they were stuck with GCC 4.2.1 to begin with.

[deanjo]

Users may or may not be, but the FreeBSD foundation certainly isn't putting technology above licencing. That is why they were stuck with GCC 4.2.1 to begin with.

They were stuck because their efforts were being put into more worthy efforts.

[pingufunkybeat]

The small performance difference between GCC and Clang is made up for in other areas such as debugging.

Oh deanjo...

Yet when people apply the same logic to open vs. closed drivers, you go berserk... B)