GNU turns 40: Stallman's baby still not ready for prime time, but hey, there's cake Happy birthday to GNU. On September 27, there will be events in both the US and Switzerland to celebrate the 40th anniversary of the GNU Project. That day in 1983, the eternally controversial Richard Stallman announced his project to create a new operating system, recursively named GNU's Not Unix. This year, the Free Software …
You're probably right. But, if it hadn't been for that AT&T lawsuit against Berkeley, things could have been very different. As it was, that lawsuit drove people to look for alternatives that were not subject to litigation.
This is not an attempt at revisionism merely highlighting one of the major boosts to the GNU profile: it was clean room code.
A BSD derivative would never have gotten as far as Linux has done over the years, one of the reasons is the license. The GPL forces customizations and modifications into the open and brings value to other users with similar requirements. It even makes fierce competitors able to work together because they know that your competitor will also share their code. The BSD license doesn't have that. The greater focus on cleanliness of code in BSD does raise the bar for getting code accepted into the kernel.
Or maybe he meant that, since they'd be borrowing from a far more complete kernel, they would have leapfrogged early Linux. Linux 0.01 barely qualifies as functional in that it booted up... and not a whole lot else. If we imagine an alternate universe where Linus took, for example, the Minix kernel and just sort of hacked it up to do things like use 386 protected memory functions, version 0.01 would probably have looked very different.
RMS may be a dick and a generally unpleasant human being, being the OG "neckbeard", but there is no denying that he managed to have a pretty significant impact on the software world. Without him there'd be no Eric Raymond to basically put Stallman's ideas into a form that was easier for people to understand, and also not be such a hard ass about everything so that the idea of open source could take hold. There's also his work with Emacs, which may be a niche program, but has had an outsized impact.
Hat's off to you if you can find a way to work with him, but the essence of compromise is everyone being willing to change at least a little bit. Based on some of the stories I've heard about him, like literally standing behind students and dictating the code they should write when his carpel tunnel got so bad he couldn't do it himself anymore, makes me think I'd probably quickly realize I'd have a more productive conversation with a brick wall, and set off in search of the nearest brick wall.
But, again, that doesn't change the fact that he has had a pretty outsized impact on the computing world. Odds are none of us will be remembered the same way RMS will, so in that regard he's accomplished more than most.
But, again, that doesn't change the fact that he has had a pretty outsized impact on the computing world.
I would suggest that his obsession with a very narrow definition of "free" and the multiple holy wars which have been started and maintained as a result, have done more to hinder the adoption of open source software than anything else. GNU/Linux my arse.
'I would suggest that his obsession with a very narrow definition of "free" '
The adjective free to mean as in freedom, in contrast to meaning without cost, has been present in the English language for quite some time. It is not a narrow definition of Mr Stallman's invention.
It's perhaps unfortunate that in English that it has 2 different meanings (German for example conveniently has _frei_ and _kostenlos_ respectively), but it's hardly the only case and most people fluent in the language are able to differentiate.
"There are currently over 80 open source licences in use."
This maybe the case but we're talking about free software here, not open-source. Free software is a subset of open-source.
A list for example at https://www.gnu.org/licenses/license-list.html.en#GPLCompatibleLicenses is fairly long but includes old versions and specific cases (e.g. PHP License, Version 3.01 used for PHPv4).
Practically, there's a choice of GPLv3 (with AGPL and LGPL variants), BSD-3-clause, the Apache license, MIT License, and the Mozilla Public License.
Isn't the point more the irony of enforcing 'Freedom' with a licence, which by definition is controlling. That's what's meant by a narrow definition of freedom.
If you put your source code in the public domain then it's free.
By definition any licence doesn't do that, it adds restrictions. That's the point. I won't release software without one as I want people to use the output of my time as I choose.
i.e. if you believe in the ideals of the GPL you'll choose the GPL licence to restrict what can be done with your source code, such as someone taking it as the basis for software that you don't release the source code to.
It may be what you want, but it is still a narrow definition of 'Free'.
Yup, that's definitely one of the issues. But then, the fundamental problem with libertarianism is that enforcing and maintaining it requires one of the most restrictive and intrusive systems of government ever suggested.
"Free as in speech" is an easy catchphrase which appeals to the dim libertarian right, but is actually of very little use when applied to software, because government restrictions on softwaring aren't the issue. Try performing a modified version of Beckett's "Endgame" for a paying audience and see how far a claim that free speech gives you the right to modify someone else's work for profit gets you.
"Isn't the point more the irony of enforcing 'Freedom' with a licence, which by definition is controlling"
Human freedoms, such as freedom of association, freedom of the press, etc. are defined by laws; laws by their nature are to do with control. This isn't ironic,it's the way modern societies function. Feel free to argue if this is actually a good idea or not, just not here as it's not really the forum.
"If you put your source code in the public domain then it's free"
Public domain usually refers to expired copyright, occasionally due to waived rights which itself is a form of licence. Perhaps you mean publicly available at no cost.
You don't have the freedom to modify the code or distribute your changes just because the source is public unless there is explicit permission through a licence. Analogue to just because a book is published doesn't mean you can change bits and publish your version. Thus a licence can grant freedoms that wouldn't exist otherwise.
The GPL does, in contrast to some other free lsoftware licences, add the requirement that derivatives are also licensed under the GPL. There's a reasonable argument for this, similar to "you can be free as long as you don't take the freedoms of others". It is not unique in this, for example the Share-Alike variants of the creative commons licences. It is also the authors perogative to insist that you must also grant the same freedoms for derivative works to others that are being granted to you
“You don't have the freedom to modify the code or distribute your changes just because the source is public unless there is explicit permission through a licence”
I think this is fundamentally incorrect and is an indication of the toxic way software is treated. I have published quite a bit of code publicly. With no licence whatsoever. If someone wants to use that code (and that was the intention, otherwise I wouldn’t have bothered publishing it at all) then fine by me. Why should I have to produce a licence?
This exact issue arose with the wire of qmail (and lots of other stuff) - Bernstein! That’s the name. I think. He has published his code freely for years and publicly stated “do what you like with it”. But some people (OS projects etc) wouldn’t use it because it didn’t have a license!! Utterly stupid. I think in the end he attached a licence that said “do what you like with it” just to keep such people happy.
Why distribution of “free” software has to get all legal is beyond me - it’s nonsense
> With no licence whatsoever. If someone wants to use that code (and that was the intention, otherwise I wouldn’t have bothered publishing it at all) then fine by me. Why should I have to produce a licence?
Purely and simply in order to state your intent.
If you publish something, you are not giving anyone the rights to reproduce and reuse that thing - like it or not, you (or your employer, in far too many situations) hold the copyright. Unless you explicitly make a statement to allow reproduction and reuse, nobody legally can.
> But some people (OS projects etc) wouldn’t use it because it didn’t have a license!! Utterly stupid.
Hey, if you want to pull down the idea of copyright, go ahead and campaign for it. Just be prepared for a few people to disagree with you.
> If you put your source code in the public domain then it's free.
Assuming you are in a jurisdiction where you can meaningfully put it in the public domain, without first dying and waiting for some 75 to 100 years.
> By definition any licence doesn't do that, it adds restrictions
"By definition"? No, not at all. The complete opposite: a licence always removes restrictions. For example, you have licenced one or more people or corporations to reproduce your work without risk of your suing them for breach of copyright. Whether or not you are asking for money (or some other consideration) in return is all part of the licence.
GPL has some very specific features, which are distinct from free, as in free beer. This has all been explained to death time and time again, but the crucial features are that the source is available to inspect, and that derivative work are to be shared under the same terms.
If you don't want that, pick a different set of licensing terms. There is no shortage of options available.
If you want a license that says do-whatever-you-want with your IP, that's fine. Don't come back crying if/when someone takes advantage of that.
"There are currently over 80 open source licences in use. That doesn't help."
This sounds plausible. However when compared to the commercial software sector where there ore over 80,000 propitiatory software licences available [1] there is still plenty of room to add a few more T's and C's
And just to point out that not all commercial licences are compatible with each other either..... I remember working at a telephone company that explicitly prohibited Oracal products to be installed on the same machine as it was.
[1] Number picked for humour value - there is, to first approximation, 1 licence for each piece of commercial software that is shipped. And that licence gets updated every few weeks with ever more restrictive terms, or requirements ratchet up the slurping of personal data depending on if the commercial software is available for the exchange of money or "free to use" (meaning that the user is the product)
"like literally standing behind students and dictating the code they should write when his carpel tunnel got so bad he couldn't do it himself"
rms has never battled with CTS. He did, however have typists hired by the FSF to do some of his typing for him due to non-CTS related hand pain. Some or all of those typists were undoubtedly students. (I'm pretty sure I got that info straight from rms, but I can't for the life of me remember when, where or how. Might be on his stallman.org site somewhere ... not that I'm prone to read his ramblings, mind.)
Gut feeling? He should have stopped using EMACS and switched to vi years ago.
I had to deal with Stallman when I put him up in my apatrment. I found him obnoxious, rude and he overstayed without even asking if that was OK. He was also very unsettling around women when my local Linux User Group took him out for an evening meal.
One thing contrary to other negative things I've heard about him was that he was very clean when it came to personal hygiene. He did leave his used teabags scattered all over the kitchen worktops though...
I've met him and wouldn't want to work with him and met others who have had and largely confirmed my suspicions. I've also met Eric Raymond and he's also a dick, but he's also contriubuted a lot. This is not uncommon amongst people with the "messianic" touch. Over time only their work will remain, which is probably best.
ESR has only contributed a lot by taking other people's work and claiming credit for it. Fetchmail, his only significant software project, was a rename of popclient. His amendments to it were poorly executed and the result was full of security holes for the best part of 20 years until others took over the maintenance of it.
His revision of the Jargon File is a travesty as well. The man himself is a libertarian wingnut with a gun fetish and bizarre religious beliefs.
> Stallman's baby still not ready for prime time
Actually; upon reflection, it kinda is "ready". Think back to the situation then. If we didn't have Linux and BSD was still faffing with licensing, we would have absolutely jumped onto GNU/Hurd as it currently is now. It isn't missing too much that render it a non-starter.
So really, it is great to know that we have BSD and Hurd ready to pick up the slack if Linux ever did have the possibility to become problematic.
...Because each of these software and technology stacks have features that make them slightly superior to their peers. They also have features that make them less than ideal for some use cases. In my opinion this is where GNU and FLOSS shine: We must chose the right tool for the job. Sometimes that means forgoing FLOSS entirely and going with Windows as the desktop and Oracle (shudder) as the database. GNU didn't just create a supply chain, but it also created a production chain so we could build the tools we need when we need them. That way we could invent new tools to make our other tools more useful.
Linux will become problematic. In fact, it's already a problem, but nobody likes to admit it. 30+ million lines of code and by all accounts only two people (Linus himself and Greg Kroah-Hartman) have a deep grasp of how it all hangs together. When they retire, get Alzheimer's (they're both about 53ish) or get hit by a bus, the whole project is in even more trouble than a vast structure which doubles in size every five years already is.
And that, of course, is if Poettering doesn't succeed in destroying it from the inside first.
Personally, I'm hoping that Haiku succeeds, because BeOS was excellent.
> Mr. Pottiecoder has moved on. He now works at Microsoft.
And Microsoft would want to stop him from releasing any more destructive[1] code into Linux?
There is probably a queue of MS execs just champing at the bit, hoping to be the one to sign the release form on his next code droppings.
[1] whether you believe his work has been inspired by the devil or by the angels, you can not overlook the divisive - and hence socially destructive - impact it has had.
But as you say, "cancer".
The problem is with the programmers who assume systemd is default, and all their things that therefore depend on systemd (even when they shouldn't).
Just like pulseaudio - on non-linux systems, you often find you need to hack programs to not need it, or needlessly port pulseaudio, or some compatibility shim.
El Reg said: "A more successful example – but also still rather incomplete – is Minix 3 ..."
The public server for Minix 3 has been more or less dead since 2016 or 2017 and no apparent development since then. There has been talk by verious people about reviving it, but it doesn't seem to come to much.
[Author here]
> no apparent development since then
True. There are 2 reasons, AFAICT:
[1] It was Prof Andy Tanenbaum's personal project, and he has retired.
https://www.osnews.com/story/136174/minix-is-dead/
[2] Intel has used it very widely, in production, in its very narrow specific role... but AFAICS it hasn't contributed back upstream.
As the other comments show, this is the problem with permissive licences vs. the GPL. Minix 3 is BSD licensed; MS has technically done nothing wrong here, but as a result, despite being used by tens of millions, nothing got back upstream -- *that I know of!*
It's a real shame.
it hasn't contributed back upstream. As the other comments show, this is the problem with permissive licences
Not really. If someone doesn't want to contribute anything back, they won't.
With the GPL, you can make source changes pretty much unusable by upstream.
This was the situation with Apple and KHTML: https://en.wikipedia.org/wiki/WebKit#Split_development
Or with NeXT and Objective-C: https://en.wikipedia.org/wiki/Objective-C#Popularization_through_NeXT
Not to mention patents, Tivoization, "cloud" services (no code released to the public), some companies' flagrant license violations, etc. And let's not forget RedHat's recent removal of source repositories and restrictive RHEL license agreement.
It tends to be bad PR that forces companies to be good stewards of open source and free software, not minimal adherence to the license.
With more liberal license like the BSDv2, companies who wish to contribute have a choice, and may decide they're in a better position to contribute money upstream, while keeping code changes to themselves. The BSDs are seeing continued development with no shortage of funding, despite their liberal licenses.
Sometimes contributing your code back is just plain self-interest, as you can get others to maintain your code for you, such as the case of Paragon's NTFS driver, and any of thousands of other cases where companies have put in considerable effort to get their horrible mess of patches into shape to be accepted into the Linux kernel. Nvidia Linux drivers serve as a counter-example, showing all the effort required to stay on top of kernel changes if you don't wish to contribute your code upstream.
Then there's lots of companies that release their code as open source despite having no obligation to do so, just because they can. Though the above free public maintenance might factor into the decision in part.
And let's not forget standardization... Lots of network services whose implementations were released under very liberal license (Apache, BIND, OpenSSH, Sun NFS, bittorrent, etc.) have become widespread standards, while I can't think of one GPL licensed implementation of a network service which has been nearly so widely successful... only Rsync comes to mind as getting some little traction.
This argument against permissive BSD/MIT licences has never made sense.
Firstly, the authors don't care what you do with it, ie. they don't want to have to deal with it. Secondy, and perhaps more importantly, if you don't add much, you have no added value to sell. And, if you do invest heavily but don't contribute upstream, you're going to have more and more work to keep in sync. I've seen this in companies a few times which think they can suddenly own a piece of open source software by forking it and keeping the fork private but I've never seen this work.
It really is a shame, and is a perfect example of why BSD and its ilk are not free licenses, and should not really even be considered open source licenses.
If somebody wants to release something as public domain, they should do that, not use these pretend "open" licenses that really exist only as a way for corporations to gobble up useful code.
If you care enough to release open source, you should care enough to release REAL copyleft open source like GPL.
I haven't really even talked to RMS since the late 1980s, when he was bopping around CMU getting misled by the Mach 3.X people about how efficient their microkernel architecture was: Sys calls were really slow, but once you mapped a file, reading the next byte was fast, so they'd tell you how fast 'getc' was, since it was basically *data_pointer++; in reality, everything else was slow. IIRC they also convinced IBM to structure (a version of?) OS/2 around the same micro-kernel principles, with the result that we all know.
But despite Hurd never coming to fruition, everything else he did, including inspiring the entire Free Software movement, has been a massive success, well beyond what I, or probably anyone else who knew him back in the 1970s, could have believed possible.
My vague recollection from the late 1970s, was that he hated AT&T for refusing to make Unix(TM) free, and he decided that in retaliation, he was going to rewrite everything they did and give it away. I mean, it was obviously ridiculous -- he's just some guy (TM), and AT&T was The Death Star.
But looking back, of course, he actually did it. Linux provided the OS, of course, but the compilers, editors, and more tools than I can count all came from GNU/FSF, and no small number of them were written by RMS himself. I've known my share of awesome programmers in my day, but i can't think of anyone who cranked out as much well designed, if poorly documented, code, than RMS. Not even close. The Free Software model has taken over the world, leaving AT&T in the dust, and even convincing Microsoft to join rather than fight.
So, let's all raise a cold one in Stallman's honor.
"So, let's all raise a cold one in Stallman's honor."
I met RMS in Brazil in a security conference around the 2000's - he was ok. Good ideas on buying Loongson hardware from China, reading emails only from the terminal, and fucking with USA intel agencies.
Drinking one for him.
...
My vague recollection from the late 1970s, was that he hated AT&T for refusing to make Unix(TM) free, and he decided that in retaliation, he was going to rewrite everything they did and give it away.
...
Actually, it was because he couldn't hack the printer driver in AT&T unix in order to fix it to work with his printer.
THAT is what set off the RMS GNU revolution.
Talk about small things with big impacts ...
"Actually, it was because he couldn't hack the printer driver in AT&T unix in order to fix it to work with his printer."
Actually, the problem was that he COULD fix the print driver, but Ma Bell's lawyers said it wasn't allowed anymore. Which made zero sense in his mind, and the minds of a lot of other people. GNU was one result. The current BSDs are another.
"OSX came from NextStep which started with a Mach microkernel, didn't it?"
The Apple kernel is (open source) Darwin which always I understood more or less as a BSD 4.4 (FreeBSD) personality on top of a Mach microkernel.
Quoting from https://developer.apple.com/library/archive/documentation/Darwin/Conceptual/KernelProgramming/About/About.html
"Up-to-date versions of the Mach 3 APIs that OS X provides are described in the Mach API reference in the kernel sources. The kernel sources can be found in the xnu project on http://kernel.macosforge.org/."
Would imply there is a fair bit of Mach 3 lurking in there. The DEC OSF/1 kernel had a similar genesis (from memory) but from Mach 2.5.
Not that Mach was directly related to RMS or MIT (was CMU) or to GNU other than the HURD kernel.
If we ostracised (cancelled) all the obnoxious pricks (dicks) in IT and computer science there would be sweet fanny left. Ditto for academia in general, politics and most other arenas.
They really only kept the OpenStep API and some of the design elements on the UI like the dock. Otherwise, OS X is fundamentally FreeBSD under the hood. But I swear around the 10.3 or 10.4 days I remember hearing about how they moved something like the printer subsystem back into the main kernel and vaguely remember hearing about similar things being moved back into the kernel instead of being separated out as the whole microkernel design would dictate with other versions. I fully admit I wasn't paying close attention at the time and my memory is fuzzy, so I could be conflating it with something else.
[Author here]
> Otherwise, OS X is fundamentally FreeBSD under the hood.
No, it isn't, but I often hear this.
FreeBSD has a conventional monolithic Unix kernel.
OS X has the XNU kernel, which is based on Mach but is no longer a microkernel, because it has an in-kernel "Unix server" derived from FreeBSD code, thus violating the core design principle of servers running in user space, communicating via IPC.
It's not a traditional monolithic kernel but it's not very micro either.
Much of the userland is BSD-derived, yes, but it doesn't use the BSD init (it has launchd instead), or the BSD system of config files in `/etc` (it uses `netinfo` instead), or an X server (it has Quartz instead). It's not even similar to BSD; it's its own thing, profoundly different, but with some superficially similar commands.
The print subsystem is CUPS, which Apple owns.
https://www.cups.org/
And of course the original Mach kernel started life as 4.2BSD, but re-written with the message passing concepts originally experimented with in the Accent kernel, also from CMU. This allowed Mach to use the tool chain and userland from BSD pretty much unchanged, and also port it to various processor architectures pretty easily. I played with all the above throughout the '80s as an interesting side diversion to what I was doing with BSD.
Fun times.
MacOS is still a lot more like BSD than Apple would like anyone to believe. It's just that it's run by another bunch of people who are scared of the shell. As a result, Apple has repeatedly had to reinvent the wheel of software releases, which is one of the reasons why it has such trouble providing patches at short notice.
From the article:
It is arguable that in the sort of narrow, specific sense that Stallman himself tends to favor, the GNU Project failed. There isn't a complete, working GNU OS. An operating system is a stack of components, from the visible user-facing stuff to the kernel, and the GNU kernel, the also recursively named Hurd, is still incomplete and not ready for daily use, even after all this time.
It's not really fair to characterize HURD as an unfinished project, because nobody's really expecting to finish it anymore. The FSF's own tediously comprehensive GNU/Linux FAQ and/or harangue even admits:
We expected to release the GNU system packaged for installation, but this plan was overtaken by events: in 1992 others were already packaging GNU variants containing Linux. Starting in 1993 we sponsored an effort to make a better and freer GNU/Linux distribution, called Debian GNU/Linux. [...]The GNU Hurd kernel never became sufficiently ready; we only recommend it to those interested in working on it.
Their user-facing recommendations exclusively promote use of the GNU system on top of the Linux kernel.
HURD is only still a thing because the entire kernel project became a manifestation of the sunk-cost fallacy. The FSF continues to develop a dead-end microkernel that will never be usable, simply because "Given the years of work we had already put into the Hurd, we decided to finish it rather than throw them away."
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
