Clang amounts to 6,645 lines of code and 868,881 lines of code in total. <- that makes no sense.
Phoronix: Apple's Dominative LLVM, Clang Statistics
After sharing GCC development statistics yesterday for this Free Software Foundation code compiler that's amassed to over seven million lines of code in 25 years, here are some development stats surrounding LLVM and the Clang C/C++ compiler...
http://www.phoronix.com/vr.php?view=MTE2MDk
Clang amounts to 6,645 lines of code and 868,881 lines of code in total. <- that makes no sense.
What exactly is it that llvm dominates? Open64? Because it certainly doesn't dominate gcc, neither in SLOC nor in performance.
Also, LLVM is not "Apple's" any more than Mesa is "VMware's" or GNOME is "Red Hat's". LLVM isn't productized by Apple at all.
OTOH there are open source products which exist that are validly assigned a brand name. Phoronix OpenBenchmarking.org? Sure. Oracle VirtualBox? Sure. GNU GCC? Yep. Nokia/Digia Qt? Yeah. But it's actually a point to Apple's credit that they don't smear their branding or proprietary licenses all over LLVM. It's basically as neutral a project as anything at freedesktop.org. It isn't commercialized; it isn't packaged and sold under a proprietary / commercial support license; it's just "pure tech".
Besides, LLVM has enough interesting applications in graphics drivers and such that there are way more companies with a big stake in LLVM than just Apple. Even AMD is starting to get a piece of the pie, and sooner or later their LLVM backend will make it into some sort of upstream repository.
You're actually doing Apple a disservice by saying "Apple's LLVM" when the only thing about it that makes it Apple's is that a significant portion of the commits are from them (but I suspect that as time goes on, other major contributors will pop up and make Apple's contributions less of an outlier and more of part of a four or five-company cabal). Apple's OS X Mountain Lion, sure. But not LLVM.
This post looks like it's written in a hurry.
I suppose that the title is meant to imply that apple dominates in the number of contributions, but it seems that that wasn't very clear (see comment above).
"So far in 2012 there have been 5,090 commits. Last year to LLVM there were 10,358 commits while the high points for Mesa were in 2009 and 2010 with 12,014 and 13,263 commits, respectively."
Why compare to mesa ? Should that be GCC ?
"Clang amounts to 6,645 lines of code and 868,881 lines of code in total." See comment above.
"There's been 38,434 Clang commits from 170 different developers." Note that 'commits' is plural, so "There have been ..."
There is a lot of interesting data in this post, it is unfortunate that it is written like this. Given the small amount of text in the post, it is quite an accomplishment to get so many errors into it.
And it's gonna take another 7-8 milions of code until it's got the performance of gcc....nothing to see here move along, apple is shitty as usual....
Sure LLVM isn't "Apples" in the sense of property but it is Apples in the way they support it. they support it in a number of ways from employing a number of developers, to hosting meetings and other ways. It is actually to Apples credit that they have remained neutral and have encouraged the very wide adoption we now see. I honestly don't think LLVM and family would have seen the rapid increases in interest and support without Apples support.
Very true, it isn't an Apple product in the commercial sense.
To be honest I sometimes wish that Apple would support it a bit more publicly. If nothing else it would get the message out that Apple does support open source. Frankly Apple employes a large number of developer contributing to open source projects that seem to be ignored by the open source / GPL crowd.
it is interesting to say the least that AMD has gone all in with respect to OpenCL and LLVM. Still one shouldn't underestimate Apples role in getting LLVM to the state it is which is leading to wide adoption by AMD and others.OTOH there are open source products which exist that are validly assigned a brand name. Phoronix OpenBenchmarking.org? Sure. Oracle VirtualBox? Sure. GNU GCC? Yep. Nokia/Digia Qt? Yeah. But it's actually a point to Apple's credit that they don't smear their branding or proprietary licenses all over LLVM. It's basically as neutral a project as anything at freedesktop.org. It isn't commercialized; it isn't packaged and sold under a proprietary / commercial support license; it's just "pure tech".
Besides, LLVM has enough interesting applications in graphics drivers and such that there are way more companies with a big stake in LLVM than just Apple. Even AMD is starting to get a piece of the pie, and sooner or later their LLVM backend will make it into some sort of upstream repository.
While I agree it is a bit of a disservice to call it "Apple's LLVM" we can't underestimate their contribution. If anything it seems like Apple has gone out of its way to minimize their association with LLVM. That is they seem to be trying hard not be overbearing in the development process.You're actually doing Apple a disservice by saying "Apple's LLVM" when the only thing about it that makes it Apple's is that a significant portion of the commits are from them (but I suspect that as time goes on, other major contributors will pop up and make Apple's contributions less of an outlier and more of part of a four or five-company cabal). Apple's OS X Mountain Lion, sure. But not LLVM.
First off LLVM is not CLang. Second if you look at how far CLang has come in the few years it has existed you should be amazed that it does as well as it does. Further the performance of generated code really isn't that bad. Sure one can find glaring examples of poor code relative to the most recent GCCs but for the most part they are neck and neck. No other compiler this new on the market has done as well.
Most importantly this is a brand new suite of tools that is open source. GCC is only open in name.