This remark appears to be a tiny side-note in a larger "State of Free Software" keynote talk, and makes Moglen look like a crazy person. I blame the title of the HN post, not Moglen.
According to wikipedia, the LLVM is licensed using the University of Illinois/NCSA Open Source License. http://en.wikipedia.org/wiki/University_of_Illinois/NCSA_Ope...
As stated in that entry, "Source code under the NCSA license can be incorporated into proprietary products without the reciprocity requirements that copyleft free software licenses raise. The license is compatible with all versions of the GNU General Public License."
Can someone smarter than me explain how this license undermines freedom? Or how the LLVM undermines freedom despite this license? I'm having a hard time connecting the dots.
Maybe I'm getting old, senile and cranky, but thing's didn't use to be this kooky in the open source movement, did they? I mean, there was always the political component to it, the strife for influence, the "stick it to the man" thing, but things didn't use to get this out of hand - smearing another open source project, simply because it's kicking your ass while having a more liberal license.
Moglen doesn't think non-copyleft software is a crime. The evidence suggests that he thinks non-copyleft software is bad for free software in the long run, because it has the effect of enhancing non-free software along with free software.
Anybody can observe that over the long run, if most of the best work in open software is done without copyleft, copyleft-protected software will suffer; free software will have one major contributor (people writing free software), and nonfree software will have two (huge companies and people writing most of the best work in free software).
It's not complicated. It's a reasonable perspective. Disagree with it all you want (and I do), but it doesn't deserve to be mocked.
But I don't think that applies. I've seen some purely technical complaints about the design of GCC. If LLVM were GPL, there would still be reasons for it to exist.
The other way to see this is that BSD allows programmers to enjoy the freedoms they got from fellow upstream developers while they take away the freedoms of their users.
http://www.faqs.org/patents/app/20100153929
Every JavaScript frontend for generating IR would violate this patent (they even mention LLVM IR explicitly in this patent). Essentially Apple just patented the UCSD p-System (nothing new here and obviously plenty of prior art so this patent is totally invalid due to prior art).
The patent has only recently been published and harms anyone considering using LLVM as an intermediate between JavaScript and back-end machine specific executables.
So much for Apple not abusing its contribution to LLVM (and in particular not abusing the freedom of downstream users).
BTW: I can see this patent being used to enforce a silo of control of using LLVM for any inbrowser HTML5+JavaScript engines (i.e. a means for optimisation).
The FSF explicitly doesn't care about whether you can make money selling lines of source code or bits in an installer. If Oracle's AGPL3 move bankrupted a startup selling a derivative of MySQL, in the FSF's worldview that BK would have been the startup's fault for building a business model that depended on restricting user rights.
I don't agree with this worldview, but I don't think it's hard to defend.
Because of the way MySQL works, that's almost certainly not true. But stipulate that it is for a second, and, what's your point? Stallman agrees with Oracle here. Web software companies are ripping other developers off by building apps derived in part from free software but keeping their own code closed.
I don't share Stallman's outlook, but I see where it's coming from.
Free software is software cannot be killed or highjacked.
Perhaps the threshold for downvoting should be a function of karma rank instead of an absolute number. Perhaps it could be reserved to the top 90%.
I did some searching, and this isn't the first time he's claimed LLVM is a threat to free software: http://digifesto.com/tag/eben-moglen/
I don't agree with him, but he's making a remark consistent with his worldview. And it's hardly outside the mainstream to target Apple for violating programmer freedom.
However, by making it sound as if a concern about LLVM was the theme of his talk, you've made him sound crazy. That's unfair and lowers the level of discourse at HN.
Moglen is making it sound as if these technical benefits weren't a factor--Apple is supporting LLVM just to undermine free software.
First of all, he can't give any example of them doing anything with LLVM that actually undermines free software. He's just speculating that they could do so in the future.
Second, there is no rational basis for claiming that this hypothetical future ability to undermine free software is the sole (or even a major) reason Apple supports LLVM.
Moglen is just spreading FUD, pure and simple.
Also, how would you suggest that the submission title be worded?
Elbowing GCC out of the dominant position in the market with a non-GPL'd component --- one that can easily be "taken private" with enhanced, patent-encumbered, closed-source variants --- is not a win for the FSF. No doubt the FSF thinks its important that the most widely-used C compiler is Free Software.
For the N thousandth time, I don't agree with Moglen, and if Moglen had given a talk titled "LLVM is a Serious Threat To Free Software", I'd think he was a crazy person. But he didn't. He made an offhand comment that totally fits with the FSF's worldview.
You've now spent far more time talking about how crazy Moglen is than Moglen spent talking about LLVM.
The appropriate title for your submission is "Eben Moglen on The State Of Free Software", since that's the title of his talk. Don't editorialize in the title; just write a comment to make your point.
The appropriate title for your submission is "Eben Moglen on
The State Of Free Software", since that's the title of his
talk. Don't editorialize in the title; just write a comment
to make your point.
However, I was submitting to promote discussion of one particular point from his talk. If I was editorializing, I would have used a title like "Moglen makes unsupported accusation against Apple" or something like that, rather than a simple, accurate conveyance of what he said.
I also found his remarks on MySQL and Oracle interesting. If someone wants to discuss them, I would see no problem with them submitting a link to his talk with a title calling attention to those specific remarks, so as to promote discussion of that particular part of the talk.
If the article you've submitted has 5,500 words in it, and the real title of that article (a) doesn't include the idea in your title and (b) accurately sums up the article, and you have to add a comment like "see minute 35:31 to see what I'm talking about" to have the title make sense, then you may have editorialized the title.
editorialize |ˌɛdəˈtɔriəˌlaɪz| verb [ intrans. ]
(of a newspaper, editor, or broadcasting organization) make comments
or express opinions rather than just report the news.
• offer one's opinion, as if in an editorial.
Alternatively, could you explain exactly how one should submit for discussion something that appeared in a longer article that discussed many unrelated things?
Which is not to say that LLVM has no technical advantages over GCC.
It has everything to do with wanting to use the same code in the IDE as the compiler and not wanting to GPL their development tools. Never mind the additional tools we are likely to see as clang and it associated projects mature.