On Fri, Dec 29, 2000 at 09:34:22AM +0200, Stanislav Malyshev a.k.a Frodo wrote:
> AS>> That's a common misconception. It should have been obvious, but
> AS>> somehow never is, that no amount of licensing trickery can make one
> AS>> program be considered a derivative work of an unrelated program. And
>
> See, this is an official position of RMS. I have quotes from him
> personally saying this. So you can tell it that way or this way but we
> both know that GPL is what RMS wants it to be, and if you try to argue,
> RMS will just force you into his position - he has enough supporters and
> public weight to do this.
>
> AS>> you link them together. What the GPL is saying that you cannot
> AS>> distribute someone's GPLed code as a part of a larger program unless
> AS>> you give users at least the same rights they would get were the
>
> Not only. You cannot also distribute "derived work", where "derived
> work" includes code made to be interoperable with GPL code. Yes, I know
> this sounds insane and goes beyond the scope of the copyright, but this is
> an official position from RMS.
Have you actually READ the GPL? It does not define "derived work"
anywhere, leaving that to copyright law. RMS has said as much, too.
> AS>> Also, nowhere does the GPL limit "interoperability". Only use code as
>
> I also thought so. But RMS says it does. Here is RMS quote about product
> that is designed to work with GPL readline library (it has no line of
> readline code, except including headers), and distributed in open-source
> with non-GPL license:
>
> ==quote==
> Richard Stallman wrote:
>
> That you don't distribute binaries does not change the fact that your
> source code is designed to include Readline in the program. You
> cannot do that, now that your license is incompatible with the GPL.
> ==end quote==
>
> It is that simple.
It is quite OK to interoperate with GPLed code if you do not make it a
part of your program. For example: MySQL in their pre-GPL days had
"mysqlclient" which included readline and communicated with "mysqld"
through a socket. I don't remember if mysqlclient was GPLed, but if it
were, the MySQL authors would have been perfectly within their rights.
Read the quote: if the program is actually including readline
(statically or dynamically) then readline becomes a part of it. We are
not talking about interoperability alone anymore, we are talking about
actual inclusion of binary code.
> AS>> part of another program. Whether or not you consider an executable and
> AS>> a dynamic library accompanying it two parts of the same program or not
> AS>> is a different issue. GPLv3 intends to clarify this and other points.
>
> I hope so, but still - GPL is what RMS says it is. And he disagrees with
> you.
I do think a program's libraries should be considered part of the
program. Or else, if someone wanted to turn GIMP into a proprietary
product, all they would need to do would to be recompile it as a
library and export all symbols.
> AS>> caused companies to become worryingly paranoid. All this has literally
> AS>> NOTHING to do with the licensing of code. The only reason why
>
> Well, I think it does. When some licensing scheme is advertized as "free",
> and you see that people avoid code by this scheme like plague -
> something's fishy here.
There sure is, but it isn't the GPL. All these companies's lawyers are
seeing is a chance for someone to sue a deep-pocketed software house,
assume that this someone therefore necessarily will, and so they are
put up policies.
Following the "never trust an open-eyed kisser" principle I'd be on my
guard when dealing with such companies.
> BTW: /usr/doc/gnome-libs-.../COPYING on my system has a perfect copy of
> GPL2, not LGPL. Am I missing something? Or you meant GTK, which _is_ LGPL?
Yes, I meant GTK :)
> AS>> exactly is the reason for the GPL. I can't see what your problem is
> AS>> here, except for the linguistic definition of "free".
>
> Yeah, exactly this - it's not "free".
Everyone has their own definitions.
> AS>> (I don't know what projects you are referring to that had to reinvent
> AS>> the wheel, but I gather they were either misinformed about the GPL's
>
> All BSD's had to rewrite readline library to get rid of GPL. That's only
> one example.
>
> AS>> restrictions, or specifically wanted their project to be forked into
> AS>> proprietary products, like the BSDs do.)
>
> Here you lost me. I know only one product of current BSDs to be
> commercialized - BSDI. And no spawns of Net*, Open*, etc.
One of the major goals of BSD is to raise the software quality bar for
proprietary products by letting vendors base such code on BSD and add
proprietary code to it[1]. I find BSD's goal admirable but it does not
settle well with GPL software author's goals, who do NOT want their
software used in proprietary products.
If a project's license gives users all rights the GPL gives, there is
no problem using readline. But you won't be able to take the project
and turn it into a proprietary one without taking out the GPLed code
first. If this is the case, BSD had to reimplement readline so that it
can be used in proprietary products.
This is different from the case of programs whose license actually
have restrictions NOT in the GPL. AFAIK FreeBSD no longer does.
See SunOS, NeXT, MacOS X for examples of complete systems based on
BSD, and numerous smaller parts of code used by proprietary systems,
e.g. sockets.
- Adi Stav
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
