RMS Feels There's "A Systematic Effort To Attack GNU Packages"

[entropy]

You need to understand RMS' thinking process here: BSD license is not free because it allows people to include it into proprietary software and this software takes the freedom from people, no matter how good it might be. The GPL ensures that software is free because the source code must be available to everybody.
So, if one major program like GDB might get threatened by another program which is not under the GPL, RMS is worried of course. This is exactly why he does not like LLVM at all.
In his opinion it is better to be unemployed than to develop non free software. He said so in a talk. Think about that one for a minute, then you see the big picture RMS is looking at and living every day.

Our world is not ready for his full set of ideas (which makes it hard to agree with RMS) but I respect what he says, because he really does care about freedom and if he doesn't preach it, then nobody will and GNU would certainly have some trouble.

If you just use free software in sense of free beer and don't care about the licenses and ideas behind them, then of course this must be utter bullshit.

Thank you!

Finally a post more meaningful than the usual hate speech.

This might be one of the talks you are referring to:

"Freedom in your computer and in the net" (31C3)

http://media.ccc.de/browse/congress/2014/31c3_-_6123_-_en_-_saal_1_-_201412291130_-_freedom_in_your_computer_and_in_the_net_-_richard_stallman.html#video

One personal note: I have to admit, I wasn't aware of the major difference between "Free Software" and "Open Source",
but what RMS says in that talk makes perfectly sense to me. In particular when it comes to Apple/LLVx.

[Luke_Wolf]

Unfortunately there's still too much reliance on GNU stuff. We still need GCC for building tons of applications (especially the upstream kernel, which is still not fully Clang-ready) and asking distributions to stop using glibc / libstdc++ in favor of libc / libc++ is a (very) tall order.

Unfortunately yeah, and the BSDs don't support my hardware yet so... between a rock and a hard place on that front, and I'm just going to have to wait. But hopefully someday soon I can have a GNU-Free system, so that I no longer have to deal with software controlled by a guy I really don't trust.

I think Theo de Raadt is as extreme as RMS in license matters.

The difference is that Theo actually remains productive to the OSS ecosystem, by being that asshole who rips out core projects that are being run poorly and rewriting/refactoring them away from crap nobody wants to touch, over to a good design, that actually promotes contributions. Further while like Stallman the guy is obsessed, he's obsessed about security as opposed to some inane notion about user freedom, which means that I can actually trust what that particular obsessed asshole creates.

[ssokolow]

I wish there was a "systematic effort to attack GNU packages". Maybe musl-libc would have attained full glibc ABI compatibility by now.

The sooner we can produce a workable, ABI-compatible distro without a single GNU component, the sooner this "Call it GNU/Linux!" nonsense can die. (Because, after all, people love to say long, awkward names and X11 being a bigger part of the codebase and desktop API than GNU is irrelevant.)

[Sonadow]

Unfortunately yeah, and the BSDs don't support my hardware yet so... between a rock and a hard place on that front, and I'm just going to have to wait. But hopefully someday soon I can have a GNU-Free system, so that I no longer have to deal with software controlled by a guy I really don't trust.

Y'know, at this point of time even Windows and OS X are better choices. At least they are governed by people who know how to cater to users and developers alike.

Anyway, if it is of any consolation, LLVM / Clang are already firmly entrenched as the No.2 in the compiler space. Most of the OpenGL / Mesa drivers in Linux require LLVM and this is not going to change. GCC cannot and will not be able to provide what LLVM does in this area, and nobody here (least of all me) wants to go back to the days of shitty graphical performance with crappy (or non-existant) drivers.

Furthermore, if you use Chrome (or Chromium) as your browser, Clang effectively supersedes GCC, especially if you roll your own build. Performing a checkout of the Chromium code automatically pulls in a local copy of Clang which is used by default in the buildscripts to compile Chromium. Even a non-programmer such as me finds the output and error log (if they occur) much more comprehensible as opposed to a few years back where Chromium used the system-installed GCC as the compiler.

And as an added bonus, I have found that ever since Google started bundling Clang with the source code checkout (which is very recent), Chromium builds much faster as opposed to GCC. Go figure

[Sonadow]

Just had a look at the mailing list; seems like Stefan has (very politely) told RMS to sod off with his opinions because he has no intention of entertaining them:

Take your time. But whatever you find out is irrelevant to whether or not the Emacs maintainer will accept LLVM support into gud.el, at least as long as I'm the maintainer.

Stefan

[justmy2cents]

So much about "the GPL is here to give you freedom". The obvious answer to this would be to make GDB better, not to shut down LLDB support.

you seem to completely misunderstand GPL and confuse it with BSD like freedom

BSD major target of freedom is user. user can do anything, even produce closed source competitive modifications.

GPL major target of freedom is software and with that user as long as he doesn't break freedom to software. as soon as user would try to break the rules, he lost the right to use it

[JS987]

I wish there was a "systematic effort to attack GNU packages". Maybe musl-libc would have attained full glibc ABI compatibility by now.

C library should be removed in order to system would be relatively secure and wouldn't be full of security bugs like buffer overflows.

[computerquip]

RMS: "Herm... I'm seeing patterns that might possibly mean that GPL licensing is being undermined in the long-term. Can you wait so I can look into that and make sure that's not the intention?"

Damn Near Everyone: "Fuck off."

I agree with RMS. He's been friendly in his concerns and asked (did not demand) for some time to look into a potential issue that he may or may not be paranoid about. And people were being complete dicks to him for really poor reasons.

Stallman, who's simply worried about the foundation that he almost solely created, probably feels betrayed that nobody else is as worried as him on the matter. I empathize to RMS on this matter, regardless of my opinion concerning GNU in general.

[bridgman]

Just to throw a wrench in the discussion...

The licensing debate keeps getting framed as "totally permissive vs GPL" where the totally permissive license is positioned as evil because it allows project code to be "taken proprietary". That's missing the point IMO -- it's not the ability to make private copies of LLVM itself that's attractive, it's the ability to build portions of LLVM alongside proprietary code and ship the combination in binary form that is attractive.

From a licensing perspective, IMO the world needs a good middle ground copyleft license, call it "corporate GPL". LGPL is close but the requirement to allow replacing project code with a newer version at any time (ie dynamic linking or supplying object files & toolchain to rebuild the binary) adds more complications than most legal departments can live with.

One big thing that has happened in the open source world recently (which might not have been noticed) is that many large companies are coming to understand that pushing their changes back upstream is worth doing not because the license forces you to do so, but because doing so means that your changes come back to you next time you refresh the project code and you don't have to do yet another stinkin' merge every time. Most of our LLVM changes go back into the public repos already, even though there is no license requirement driving it, and the percentage of code pushed back is continuing to grow.

If LLVM had a license which required source changes to be made available, but which did not require all of the SURROUNDING source code to be published, I don't think that would impact its usage. In a perverse way this is another DRM scenario, where mechanisms put in place to avoid abuse (the viral aspects of GPL were added with the best of intentions, ie to ensure that added features & extensions to a GPL project were also covered by the GPL) ended up having unintended consequences, eg using gcc as a shader compiler in a proprietary graphics driver would require the whole graphics driver (and possibly more) to be released under the GPL. I doubt that was the original intention.

The real solution here IMO is some kind of "CGPL" license, where changes to project code must be published back, extensions and additions that fall within the project scope must be pushed back, but static linking of project code into an obviously unrelated project does not require source for that other project to be published.

[mark_]

C library should be removed in order to system would be relatively secure and wouldn't be full of security bugs like buffer overflows.

you don't need glibc with systemd anyways