Google ? [LWN.net]

Posted May 23, 2007 19:38 UTC (Wed) by tseaver (subscriber, #1544) [Link]

The wish of some (e.g., in the "Affero" license) to extend that definition
to include situations not covered by copyright law are likely to end
up unenforceable, at best.

Posted May 23, 2007 23:32 UTC (Wed) by MathFox (guest, #6104) [Link]

Copyright law also covers "public performance" and a case can be made that running software on a publicly available computer (for instance a webserver) constitutes public performance of the program.
IANAL and I know that the details of the law vary by country... Affero might be stretching the GPL a bit, certainly not recommended for every application, but it is founded on the letter of the law.

Posted May 23, 2007 23:56 UTC (Wed) by xoddam (subscriber, #2322) [Link]

That's true, but copyright law (in some countries) covers things other
than distribution; for instance the right to prepare modified works or (as
a previous reply mentions) to perform the work to a public audience. The
point is that it is illegal (in those jurisdictions) to make "in house"
modifications except as allowed by the copyright licence (and by "fair
use", if such a right also exists in that country).

The GPL v2 gives blanket permission to modify the code as the licensee
pleases. The Affero Public Licence withholds that permission in specific
circumstances, namely it prohibits the removal of a prominent "download
the version of the code which provides this service" button from a public
web service.

The new notion of "propagation" in the most recent draft of the GPL v3
allows the extension of copyleft protection into areas not normally
considered by copyright law, by selectively withholding permission to do
things which *are* the provenance of copyright. This applies, in
particular, to the GPLv3's patent and anti-DRM provisions, and to the
no-more-MS-Novell-deals clause.

It might apply equally effectively if "propagation" were defined also to
include providing a service to the public based on the covered work. The
current GPL v3 draft does not do this; a later version of the Affero
licence probably will.

If copyright gives holders the right to demand payment per copy or for a
licence to prepare derivative works, it certainly gives them the right to
impose conditions on the way in which those actions are permitted.

Nevertheless I don't think applying those terms to the GNU body of code
would be sufficient to force a company the size of Google to publish its
in-house code. I am certain they are very well aware of what code they
own and what they have licensed from elsewhere, and of the terms of each
and every license.

Such a move might be so counterproductive as to move Google from the "free
software is a good thing and I'm pleased to contribute even if not all my
products are free software" camp into the "FSF are communists who want me
to give away my golden goose" camp.

Posted May 31, 2007 12:35 UTC (Thu) by zotz (guest, #26117) [Link]

I don't see why people seem to thing this. Can you explain the thinking for me?

I can deny the right to copy my work even in house unless you agree to the conditions set in the license. Yes? No?

all the best,

drew

Google ?

Posted May 24, 2007 2:30 UTC (Thu) by lysse (guest, #3190) [Link]

> It hasn't changed, but it was a considered change for the GPLv3 that provided web services and the like should count as a form of distribution.

By that standard, then if I provide someone with a shell account on my machine, then I am distributing every program they can access to them; even if I only provide execute access to the programs, I would still have to provide them with a means of downloading all source code. I can't see any sensible way of defining "distribution" that would include web services but exclude this; one might draw a distinction between public distribution and limited distribution, but that would make the desirable qualities of the GPL easily circumventable.

So such a restriction would have to be defined as a restriction on use, rather than on redistribution... and then you'd have to define "use" in a way that includes "present to the world as a web app" for Perl scripts, but excludes "compile a program" for gcc, or "produce a Postscript file from this text file" for groff - or even "run this script" for Perl itself - and you'd have to be absolutely sure your definition was both precise and enforceable.

Er, eek?

Google ?

Posted May 31, 2007 12:39 UTC (Thu) by zotz (guest, #26117) [Link]

"By that standard, then if I provide someone with a shell account on my machine"

Could be. Anyone agree? Disagree?

What about putting together a system, hardware and modified software, and renting it out.

all the best,

drew

Google ?

Posted May 24, 2007 9:17 UTC (Thu) by anselm (subscriber, #2796) [Link]

Let's stipulate for a moment that you are right about the »spirit« of the GPL, as applied to web services.

The question here is how far down the stack this should extend. Presumably Google are running their own proprietary search engine code, on which the »spirit of the GPL« has no bearing, on top of a possibly-patched Linux kernel. Should they be morally obligated to publish their Linux kernel patches just because the machine in question is running (or even merely talking to another machine running) a web server that is accessible by the public?

Google ?

Posted May 24, 2007 22:57 UTC (Thu) by giraffedata (subscriber, #1954) [Link]

No, something changed.

The GPL didn't change, and legal requirements didn't change, but the thing that Stallman sought to regulate with GPL has changed. Much of the computer use that formerly happened via distribution of software now happens by offering a service on the Internet. For example, this year I filed my income tax using a web service, whereas in the past I would pick up a CD at a store. And it looks like that trend will continue. Essentially, we're moving from the Microsoft era to the Google era.

So to continue the freedoms GPL hoped to give to computer users, there needs to be something new.