Stallman calls LLVM "A Terrible Setback" | Hacker News Reader

Stallman calls LLVM "A Terrible Setback"

phoronix.com

5 pointsmindcrime12 comments

So, is the following an accurate description of the situation? 1. gcc is kludged to preclude use in non-free software. 2. llvm's design is modular and doesn't preclude such use. 3. In RMS's opinion, a politically motivated kludge is to be preferred.

If you believe in software freedom for all users, then it is immoral to distribute proprietary software or enable others to do so. In that case it's not a kludge, it's a design feature to make it harder for others (eg, "software hoarders") to be immoral, and your points are incorrect.

If you believe in software freedom for all developers, so they can do whatever they want to do with your code, then yes the gcc design is horrible and your points are correct.

The choice of viewpoint is based on one's politics and ethics, even if isn't explicitly stated. You apparently don't believe in software freedom for all users, so you interpret RMS's objections as being "politically motivated."

But by the same token, clang has made a deliberate choice to not protect user freedom, which can just as easily be interpreted as a politically motivated decision.

Hmmm... then what distinguishes that design feature from DRM?

Ummm, lots of things? To start with, access to the source code, the legal ability to modify and redistribute said code, and no requirement for permission renewals from someone else.

What gives you the idea that this is anything like DRM?

Given the long history of spin-offs like gcc-xml (which started in 2000), there's nothing which prevents you from adding this design yourself. This is a social contract issue, not a legal one.

Bias much? An accurate description would be "gcc is licensed such that plug-ins must be at least as freely licensed as gcc is (GPL3) and LLVM is licensed such that proprietary modules can plug in". If that's a "kludge", then it's a pretty "interesting" use of the word, and one that can be applied to an awful lot of software. Further, it's philosophically motivated, not politically. Whether or not you agree with it is a different, and frankly irrelevant issue.

I was first introduced to the free software philosophy when I saw Dr. Stallman at a conference in a local university. Even though I often find his positions a little too "cut-throat", I must admit he's very stable in his argumentation.

"The existence of LLVM is a terrible setback for our community precisely because it is not copylefted and can be used as the basis for nonfree compilers -- so that all contribution to LLVM directly helps proprietary software as much as it helps us."

As unpopular as it may sound, I think we need such opinions. He has a set of values and tries to defend them. While it may sometimes not be pretty, it's all part of a healthy debate.

A healthy debate requires viable alternatives, that is, LLVM. Doesn't look like a setback to me.

Stallman made the same argument about Java bytecode back when it was the cool low-level cross-platform compatibility layer.

[1] http://gcc.gnu.org/ml/gcc/2001-02/msg00895.html

There's no other way I can comment this.

The thing is, he's not actually wrong. Not IF you share the same values as Stallman, where "software freedom" is the most primal and important value / goal of all. If you don't share his values, then this statement seems absurd, of course. But he does have a point, even if one chooses to disagree with him.

(note: just playing Devil's Advocate here. I'm not a GNU/FSF hardliner. In fact, I use the ALv2 for most of my open source code.)

> The thing is, he's not actually wrong. Not IF you share the same values as Stallman, where "software freedom" is the most primal and important value / goal of all.

That's unquestionably true, if you share Stallman's particular view that the important part of "software freedom" is its negative aspect -- that is, the absence of non-Free software, not its positive aspects, the availability for use of Free software.

If you value the same freedoms the FSF embraces but prioiritize positive software freedom more than negative software freedom, its a lot less clear that a LLVM's Free-but-easier-to-use-with-or-in-non-Free-systems nature actually harms software freedom rather than enhancing it.

If you value the same freedoms the FSF embraces but prioiritize positive software freedom more than negative software freedom, its a lot less clear that a LLVM's Free-but-easier-to-use-with-or-in-non-Free-systems nature actually harms software freedom rather than enhancing it.

Agreed.

Reply on news.ycombinator.com