Should it be GNU/Linux [LWN.net]

Should it be GNU/Linux

Posted Mar 23, 2007 18:30 UTC (Fri) by landley (guest, #6789)
In reply to: Should it be GNU/Linux by man_ls
Parent article: The road to freedom in the embedded world

> IMHO it is reasonable for the FSF to ask people to use the
> name "GNU/Linux"; after all, it was Stallman who came out with the idea
> of a libre, complete operating system, and they are the only ones
> providing it...

Ok, let's go to http://www.ibiblio.org/jmaynard/ and read the bit right
above the download link:

> This site contains copies of distribution tapes, other source and
> object code libraries, and pregenerated, runnable distributions of IBM
> public domain software written for the System/360 and System/370
> mainframe computers.
>
> All of the software on this site is in the public domain. IBM, by
> corporate policy, does not assert copyright ownership of any software
> which it distributed without copyright notices. US copyright law, until
> 1978, placed such materials in the public domain.

I repeat my earlier assertion that for most of the 1970's proprietary
software wasn't even an issue on the hobbyist programmer community's
radar. When Gates did his "letter to hobbyists" in 1976 the response was
essentially to laugh it off. The Lyons book was published before anyone
knew if it was even legally possible to copyright source code, which was
made explicit by a new law passed in 1979, and whether or not that
copyright extension covered binaries wasn't settled until Apple sued
Franklin over the ROM images in its' Apple II clones in 1983. (So
between 1979 and 1983 if you wanted to be sure copyright covered your
work you made darn sure to distribute source code with prominent
copyright notices.)

Which gets us back to "the FSF wasn't being visionary, it was being
reactionary and conservative from day 1".

> Linux itself has already been replaced.

Plus existence of BSD and MacOSX and the Posix certifications of Windows
NT and OS/360...

Closer to home, Shawn Jackman built a subset of the BusyBox applets under
newlib+libgloss in 2005, running against the bare metal with no
underlying OS kernel.

What Linux pioneered was modern open source collaborative development
through the internet with release early/release often and all that jazz.
And he's pioneered scaling that mode with a driver maintainers layer
between the project's leader and the developers, and then it went to a
four-layer thing with the lieutenants (essentially subsystem maintainers
above the driver maintainers). Plus being the first project to apply a
modern distributed source control system to open source development.

The Cathedral and the Bazaar was a paper about how Linus's working style
differed from that of the FSF. (The cathedral was specifically the FSF.)
The FSF benefitted from the internet but Linux was the first development
project predicated on taking full advantage of it from day one. (The FSF
insisted on physical copyright assignments with a signature on a piece of
paper, and still do. Linus often merges over a hundred patches in a day,
that kind of bureaucracy just wouldn't work.)

(Log in to post comments)

Should it be GNU/Linux

Posted Mar 24, 2007 0:40 UTC (Sat) by man_ls (subscriber, #15091) [Link]

Ok, let's go to http://www.ibiblio.org/jmaynard/ and read the bit right above the download link:

I'm not sure I follow you, the existence of large libraries of public domain code (even complete operating systems) at a moment in time is nice, but the real problem would come when IBM and the rest changed their view on copyright and started asserting it. Stallman set out to make an OS which could not be made proprietary.

I repeat my earlier assertion that for most of the 1970's proprietary software wasn't even an issue on the hobbyist programmer community's radar.

True, that is precisely why Stallman was being visionary by being concerned before the rest of the world saw the problem of the proprietary approach everyone was taking.

Which gets us back to "the FSF wasn't being visionary, it was being reactionary and conservative from day 1".

A bit contradictory (or again I'm not following you), Stallman saw the problem not only with proprietary development, but also with public domain and BSD-style licenses. That is why he created the GPLv2 and the FSF. Why is that "reactionary"?

The Cathedral and the Bazaar was a paper about how Linus's working style differed from that of the FSF. (The cathedral was specifically the FSF.)

I have seen this assertion of yours a couple of times, and it is what made me answer this post: where do you get this impression? Even in the abstract we read:

I discuss these theories in terms of two fundamentally different development styles, the ``cathedral'' model of most of the commercial world versus the ``bazaar'' model of the Linux world.

Or this paragraph:

Perhaps this is not only the future of open-source software. No closed-source developer can match the pool of talent the Linux community can bring to bear on a problem. Very few could afford even to hire the more than 200 (1999: 600, 2000: 800) people who have contributed to fetchmail!

I don't think Raymond considers GNU as "closed-source". There are a thousand examples all over the text where "cathedral" is equated with "closed-source", "proprietary", "commercial" etc. It is true that FSF development seems to be more "cathedralicious" or centralized than other free software projects, but saying that they were the epitome of cathedral development is quite misguided IMHO.

The FSF insisted on physical copyright assignments with a signature on a piece of paper, and still do. Linus often merges over a hundred patches in a day, that kind of bureaucracy just wouldn't work.

I don't see why not. The piece of paper is just needed once per developer; after that you can contribute as much as you want. Given that developers do not change that much, that many of them come from companies (which can ease much of the paperwork) and that copyright assignment is not needed for one-line patches, it is not a significant entry barrier.

Should it be GNU/Linux

Posted Mar 24, 2007 17:48 UTC (Sat) by landley (guest, #6789) [Link]

> > Ok, let's go to http://www.ibiblio.org/jmaynard/ and read the bit
> > right above the download link:
>
> I'm not sure I follow you, the existence of large libraries of public
> domain code (even complete operating systems) at a moment in time is
> nice, but the real problem would come when IBM and the rest changed
> their view on copyright and started asserting it. Stallman set out to
> make an OS which could not be made proprietary.

IBM didn't make the existing versions proprietary, they stopped releasing
new open versions. (Keep in mind IBM didn't accept outside contributions
to their OS. The modern equivalent of this is that the sole copyright
holder were free to issue new licenses, as Sun is doing with the GPL.
But the code that was out there stayed out there.)

IBM's big change of heart was called the "Object Code Only (OCO) Policy",
issued February 8, 1983. Here's a copy of their original announcement:
http://landley.net/history/mirror/ibm/oco.html

Here's the final report of SHARE's attempt to get IBM to change its mind
(essentially giving up 5 years later):
http://www.redbug.org/dba/sharerpt/share71/s987.html

And an history article putting it in context:
http://www.itworld.com/Comp/1369/LWD000606S390/

> > I repeat my earlier assertion that for most of the 1970's proprietary
> > software wasn't even an issue on the hobbyist programmer community's
> > radar.
>
> True, that is precisely why Stallman was being visionary by being
> concerned before the rest of the world saw the problem of the
> proprietary approach everyone was taking.

Going from a world that had less than a hundred thousand computers in it
to a world that has more than a hundred million in the space of about 15
years led to some impressive culture shock on the part of the programming
community. What happened is that most of the existing programmers didn't
see alternative ways of doing things as a _threat_, because they didn't
realize they were about to be outnumbered by a factor of 1000.

> > Which gets us back to "the FSF wasn't being visionary, it was being
> > reactionary and conservative from day 1".
>
> A bit contradictory (or again I'm not following you), Stallman saw the
> problem not only with proprietary development, but also with public
> domain and BSD-style licenses. That is why he created the GPLv2 and the
> FSF. Why is that "reactionary"?

He was striving to maintain the status quo. Return to the glorius past.
The way we did it in the good old days was superior to these newfangled
professional software development businesses.

He may have been right, but that doesn't change the nature of his
actions. And how right he was when what he was originally trying to
defend was the ITS system against the MIT administration, and the rebase
to Unix only came about when the PDP-10 hardware line and
proposed "Jupiter" follow-on project were cancelled (also in 1983),
rendering ITS (written entirely in PDP-10 assembly) a clear dead-end.
His move to Unix was forced upon him when ITS died, he just wanted to
move as little as possible.

Fast forward to _today_ and people are going "oh, what great new insights
do you have for us with your keen eye for the future" when all he ever
did was prefer 1977 to 1983. I do not look for great insights from this
man, I look for clever hacks to defend ideas from the 1970's, often with
long elaborate rationalizations for things he's already made up his mind
on. (Show me the last time he _changed_ his mind due to new information.
Yeah I know, he's not inflexible, he's making a stand on principle. I
honestly thought he might take up the cause of deCSS in 2000, but ITS
couldn't play DVDs. There still isn't a GNU deCSS implementation. Not a
battle he wants to fight.)

> > The Cathedral and the Bazaar was a paper about how Linus's working
> > style differed from that of the FSF. (The cathedral was specifically
> > the FSF.)
>
> I have seen this assertion of yours a couple of times, and it is what
> made me answer this post: where do you get this impression?

From Eric Raymond directly, while editing The Art of Unix Programming
(check the intro for my name), and outright co-authoring things like the
OSI reaction paper to the SCO lawsuit, Halloween 9, and the 64-bit paper
(which he insisted on titling "world domination 201")...

It's in the book, though, if you look for it:
http://www.catb.org/~esr/writings/cathedral-bazaar/cathed...

> But few of us really thought very hard about what we were doing, or
> about what the very existence of that archive suggested about problems
> in the FSF's cathedral-building development model.

A couple times he's talked about a specific 1996 conference that Tim
O'Reiley put together (and that he, Linus, and RMS attended) where worked
it all out for the first time, seeing them next to each other and
thinking about the differences. I could ask him for details...

> I don't see why not. The piece of paper is just needed once per
> developer;

Needed at precisely the wrong time.

I touched on it here: http://lkml.org/lkml/2002/1/29/9

Most developers start out as casual contributors. A line here, a bugfix
there. The less they have invested in participating in a project's
development, the more easily discouraged they are. Needing a sign-off to
get cvs commit access is one thing, but needing a sign-off to take your
five line function? Eh, it wasn't that important anyway.

> developers do not change that much,

In a project with 1000 semi-regular contributors, the top 20 don't change
that much month to month. Call the newspapers.

The point is where do they come from? Let's look at a couple of
examples:

Con Koliavs: scheduler dude. When he got into Linux his day job was as
an Australian anesthesiologist, he started poking around Linux for fun.
First time he wandered away, two years later he tried again and got
hooked:
http://kerneltrap.org/node/465

Andrew Morton, current #2 in the development community. Only got involved
in the project in 2000, because a NIC he was using had been declared
obsolete and he sent in a patch fixing it. (Was anybody other than him
still _using_ that NIC? Dunno. He could have maintained it out of tree,
but it was easier to get it merged so it wouldn't break again.)
http://kerneltrap.org/node/10

Even the early adopters did stuff before Linux. Alan Cox used to do
Amiga stuff and MUDs, Peter Anvin first used Linux to put together
terminal servers...

Back when I read the first year or so of the linux kernel mailing list, I
collected a few interesting posts (with links back to the originals in
the archive):
http://landley.net/history/mirror/linux/1991.html
http://landley.net/history/mirror/linux/1992.html

Now ask yourself: how many of those people would have just wandered away
again if Linus had asked them to fill out paperwork as a condition of
participating?

Attracting and breaking in new developers is extremely important to the
long-term health of a project. This is why things like kernelnewbies.org
exist.

Rob

Should it be GNU/Linux

Posted Mar 26, 2007 12:58 UTC (Mon) by nix (subscriber, #2304) [Link]

`A line here, a bugfix there' do not and have never required copyright assignment papers.

Should it be GNU/Linux

Posted Mar 26, 2007 16:53 UTC (Mon) by bronson (subscriber, #4806) [Link]

Sure they did. Back in the late 90s, I wanted to submit a 15 line rather obvious bugfix to the viper elisp package. It took a snail mail letter and three weeks to clear up the copyright assignment. I almost didn't bother.

I do hope you're right and the FSF is being more realistic now. Not to long ago, though, they required papers for just about anything. I think I still have the proof in my filing cabinet!

Should it be GNU/Linux

Posted Mar 27, 2007 14:37 UTC (Tue) by nix (subscriber, #2304) [Link]

The rule-of-thumb the FSF uses is `if the fix is trivial, no assignment is required: things ten lines long or less are assumed trivial'.

It looks like you slipped in just above that limit.

(I find it very little effort to do the occasional copyright assignment dance: the *really* annoying part is getting your employer to disclaimer-of-rights forms, because even if they agree they can be *so* damn slow at it it's not true. I've had to wait >6mths for these sometimes.)

Should it be GNU/Linux

Posted Mar 26, 2007 21:55 UTC (Mon) by man_ls (subscriber, #15091) [Link]

[about RMS, and his visionary qualities]
>Why is that "reactionary"?
He was striving to maintain the status quo. Return to the glorius past.

That is like saying that Mohandas Gandhi was reactionary because he just wanted to "return to the glorious past" when the English did not rule India. True, but lopsided.

Stallman wanted not to return to a naïve past where kind people shared code, but go on to a new situation where people wrote free software because they wanted it to be free, and companies (a new concept) pooled their effort. Guess what, it works.

[again about RMS, and his cathedralicious model of development]
It's in the book, though, if you look for it:
http://www.catb.org/~esr/writings/cathedral-bazaar/cathed...

Ah, you worried me. For a moment I thought it was there, plain for all to see as a central recurrent theme and I had missed it: that instead of talking about closed, proprietary software Raymond was speaking about his old buddy Stallman all the time. But it is sort of hidden, isn't it?