GPL vs BSD Licence

[David Schwartz]

> No, they are arguing that the RESULT of the linking becomes a derived
> work, NOT the original, unlinked work itself.

	Well, if they're arguing that, they're right, but the argument is
irrelevent. If I run any executable on a Linux box, the RESULT of the
running (the state of my computer) is a derived work, NOT the original,
unexecuted work. Since nobody is distributing the derived wok, and the GPL
allows unrestricted use, why is this important?

> This is a fine line here, and one that is more of a legal argument than
> anything else.
>
> I think the idea of it goes something along these lines.  Suppose you
> distribute a non-functional program along with instructions that to
> make it functional, the user must fetch some GPL program and link
> that in.  No problem there.

	Or, of course, any other program that implemented the same interface as the
GPL program. This is the key, it's not the copyrighted expression (the
source code) of the GPL product that is being used, but it's (unprotected)
interface to the outside world.

> But then suppose you distribute the same program along with the GPL
> program and a set of instructions the user must follow to build it,
> and to get it running.  One again, no problem there.
>
> But then, suppose you distribute your program AND the GPL program AND
> when your program is run, it automatically pulls in the GPL program
> and links into it and does it's thing.  THAT I think is where they
> are arguing that your crossing the line.

	Why? You are still distributing the unmodified GPL program and you are
still only deriving from its interface, not its code.

> From one view of things, they do have a point - as your third distribution
> is in effect a way to dodge the effects of the GPL.

	No, it's not a way to dodge the effects of the GPL, it's a way not to
create a derivative work. It's no more a dodge than developing a work-alike
in a clean room is a dodge.

	I can't stand when people interpret defining the legitimate bounds of
copyright as ways to 'dodge' the law or 'steal' other people's work. (And
I'm not saying that *you* are doing that!)

	The law protects the original expression in the source code and anything
that directly flows from that. It does not protect the interface into and
out of the program, unless that also contains protected expression.

> But of course
> from a strict copyright interpretation point of view, I think that this
> is the weakest argument the FSF is making about the GPL because I think
> the history of court cases of these nature has shown that courts generally
> say that even when code is intermixed, that each entity retains it's
> own copyright.

	Exactly, so long as they are not intermixed in a way that you can't draw a
line between them. Including a header file is a tough case, because they are
more thoroughly intermixed. But linking, IMO, leaves a clear line between
the two works. And even if it didn't, the line is certainly there before the
link, and dynamic linking is use.

> > Nothing that you can do with a work after it's produced can
> > turn that work
> > into a derivative work of another work.
>
> That isn't and wasn't ever the point.  Your "you" here is the secondary
> user not the original author.  The example I gave was that the "you" was
> the original coder/copyright holder.

	You do admit that you can't link code until you've written it, right? The
copyrightable expression is the writing of the source code. Nothing that
happens after the source code is written can make the copyrightable
expression not fully original.

> > Even if the program had been designed from the beginning to
> > work with an
> > interface that was currently only implemented in a GPL'd
> > library, I would
> > still find the argument absurd (though slightly less so).

> Once again, I've said it several times before but you don't seem to
> get it, so I'll say it again.  The GPL is concerned with DISTRIBUTION
> -not- MANUFACTURE.

	The distribution is in the form of mere aggregation. You can't make one
work a derivative of another by distributing them together. The GPL only has
force on a work if it's a derivative work.

> If you design a program from the beginning that works with an interface
> only implemented in a GPL library, then there is NO problem if
> that code is NOT GPL'd, AS LONG AS you aren't distributing a compiled
> version of it that links in the GPL library.

	Why? The distribution occurs after the work is made. It can't make the work
a derivative work. And the distribution itself is of the mere aggregation of
the two works -- there is a *clear* line between them. The combined work is
only formed at execution time. You might as well argue that RedHat can't
distribute the Linux kernel in a CD set with any work that's not GPL'd
because when you install and run them, they combine with the Linux kernel
(or other libraries). Yes, they do. So what? That's *use*.

	RedHat has just as much tweaked all those products to work together as the
hypothetical linking case we're talking about here. And it still doesn't
make any of the works derivative of any of the others. Legally speaking
(grossly oversimplified), one work is a derivative work of another if the
protected expression in one work contains some significant portion of the
protected expression in another.

> If the end users -must-
> go compile in some GPL code that is perfectly fine as long as THEY
> have to consciously choose to do it, NOT YOU (or your program).

	I find that argument silly. Nothing that happens after a work is created
can change its copyright status. The GPL can only apply to a work if the
author chooses to apply it, or if that work is derivative of a GPL'd work.

> You know I said this already on Tuesday, the GPL is concerned with
> REDISTRIBUTION not COPYRIGHT.  Why do you insist in ignoring this?

	I can't comprehend what you mean. It's concerned with redistribution, not
use. But it can only cover a work if copyright law allows someone who chose
to use the GPL to have some rights over that work.

> You know, I really find this to be extremely ironic.  I am a big
> opponent of the GPL license and always have been.  But I find myself
> playing devils advocate here for the GPL because so many of you
> who should damn well know better quite obviously don't know the
> first damn thing about the GPL.

	I'm sorry, what don't I know exactly? (Are you talking about me or some
general class of GPL bashers?)

> Clue phone ringing here!  DON'T TRY ARGUING AGAINST SOMETHING YOU
> DON'T KNOW SHIT ABOUT.
>
> If you thing the GPL is a pile of poop, as I do, THEN LEARN THE DAMN
> THING BETTER THAN MOST OF THE GPL ADVOCATES OUT THERE.  Trust me,
> IT AIN'T HARD!!

	I think I know the GPL about as well as the advocates do. I don't think
it's a pile of poop, but I don't like its effects and prefer non-GPL'd
projects and works to GPL'd ones.

> 95% of the GPL advocates out there don't know the first damn thing
> about the GPL and undoubtedly have never completely read through
> the thing, MUCH LESS have read the supporting philosophy and writing
> around it.  I see GPL bigots ALL THE TIME arguing about how great
> GPL code is and how the GPL is so all fired wonderful because of this
> and that which isn't even IN the damn license!!!
>
> Read the GPL.  Read the FSF's website and RMS's writings.  Read the BSD
> copyright.  THEN if you want to go bash the GPL please do so!!!

	I have, and while I don't bash them, I disagree with many of their
interpretations because they don't seem to reflect any understanding
whatsoever of copyright law. I think part of this is because RMS often
rights about the world the way he wishes it was rather than the way it
actually is.

	Again, the GPL can only affect a work if someone with legal copyright over
that work decides to apply it. That would mean the authors/owners, unless
the work could legally be classified as a derivative work of a GPL'd work.
This is a copyright law technical term that the words of the GPL cannot
change because its legal authority to affect the work at all comes from
whether it has or hasn't that particular legal status in the first place.

	This article is reasonably on point:
http://linuxjournal.com/article.php?sid=6366

	DS