Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 16 May 2000 10:28:52 +0530
From:      Rahul Siddharthan <>
To:        David Schwartz <>
Cc:        Anatoly Vorobey <>, chat@FreeBSD.ORG
Subject:   Re: RE: Why are people against GNU? WAS Re: 5.0 already?
Message-ID:  <>
In-Reply-To: <000001bfbed5$21a44b30$>; from on Mon, May 15, 2000 at 06:21:59PM -0700
References:  <> <000001bfbed5$21a44b30$>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help
> > OK suppose you're right (which is very unlikely).  What then?  The GPL
> > still only talks of using later versions at your *option*.  You can
> > still use GPL 2 if you want.
> 	The point is that if you cover a work under the GPL, and then others modify
> and extend that work, the only entity that has any control over the
> licensing terms is the FSF. Not even the original owner can modify the
> licensing terms.

The point is, in this case the original owner is no longer responsible
for the whole work.  But he/she can still change the terms of the
original, unextended work.  This is hardly relevant.  Once the
original owner has given the work to someone else under a particular
license, he can no longer change the licensing terms, *whether it
has been extended or not*.  He has given it and cannot take it back,
new licensing terms can only take place for future distribution of
work.  Moreover, the person who received the work under the GPL can
continue to redistribute under the GPL, even without modifying, no
matter how the original owner decides to license it in the future.
This is as true of the BSD license as of the GPL.  Once you've given
it to someone with such a permissive license, you can't take it back
or change the licensing terms.

> This has, historically, created some amusing incidents. Take for example
> the Linux operating system. It was developed under the GPL from day one. It
> contains contributions from large numbers of people, all contributed under
> the GPL. Theoretically, nobody but the FSF should be able to change the
> licensing agreement. Not even the original author (Linus Torvalds) is
> supposed to be allowed to do that.
> 	But he did. He declared unilaterally the programs that just Linux kernel
> calls are not covered by the GPL. Now what is this declaration? Is it Linus'
> legal opinion?
> 	Isn't this in direct conflict with what RMS recently said on this list?
> Linux is not covered by the LGPL, and all programs that run on Linux 'link'
> to the Linux kernel as surely as any library use.
> 	Now somehow in this case, Linus seems to have gotten away with it. But I
> have no idea how he has the right to change the licensing terms on other
> people's software. As I read the GPL, only the FSF has that right.

It is true, he probably doesn't have the right and RMS has expressed
himself against it.  But I think I see your problem.  You're not
worrying that the FSF has the *power* to do anything, you're worrying
that the FSF has the *ability*.  The thing is, even if the FSF decides
to change the GPL, it's up to Linus whether or not to accept the new
GPL.  Even if someone else decides to distribute with the new GPL,
Linus can continue to distribute with the old if he chooses to, and
his licensing terms are likely to be treated as authoritative.  
So the FSF can't harm the interests of the community: the more
suitable license will prevail.

> How do you address this? Does Linus have the right to relax the GPL on all
> the contributed works in the Linux kernel? Or will RMS go on record as
> saying that running on an OS is different from linking to a library? (How
> so? Technically, they are precisely analogous.)
> Bluntly, Linus should not have applied the GPL to his work. He should have
> applied a license that had the terms he wanted to apply, and this would have
> allowed him to fix erroneous terms later. But he ceded that power to the
> FSF. Now, if someone really wanted to make a stink, I'm not sure what he
> could do but throw out all code contributed to the kernel inbetween when it
> was first released and when he clarified the license.

The only person who's likely to want to make a stink is RMS, and he
hasn't -- that is, he's spoken but not acted, and afaik he's spoken on
principles (don't encourage binary-only drivers) rather than legal
issues (the GPL doesn't allow it).  I think he realises that there's
an ambiguity in the GPL about what constitutes a derived work.  That
is, is a binary driver which relies on system calls from the kernel,
but uses no code from the kernel and is loaded only as a runtime
module, "independent and separate".  


To Unsubscribe: send mail to
with "unsubscribe freebsd-chat" in the body of the message

Want to link to this message? Use this URL: <>