APSL 2.0

Post by Jens Schmalzing
Hi,
I've just stumbled upon Apple's claim [1] that their latest version
2.0 of the Apple Public Source License [2] has been certified as free
by the FSF. Not being an expert in these matters, I am wondering if
this makes the license DFSG-free - if this was the case, I would make
efforts to re-join the Mac-on-Linux packages. Anyone care to
enlighten me?

Jens please note that I am not a Debian developer.

As I read the APSL 2.0 it is unlikely it will be considered Debian free
software. As you note, Richard Stallman ("We are grateful to Richard
Stallman for his many helpful comments in this process") and the Free
Software Foundation has approved of the new Apple Public Source License
(APSL) 2.0 as a Free Software licence:
http://www.opensource.apple.com/news/2.0-announce.html
http://www.fsf.org/philosophy/apsl.html
http://apple.slashdot.org/apple/03/08/06/1729213.shtml

Look at the APSL 2.0 definition of "Externally Deploy":

`"Externally Deploy" means: (a) to sublicense, distribute or otherwise
make Covered Code available, directly or indirectly, to anyone other
than You; and/or (b) to use Covered Code, alone or as part of a Larger
Work, in any way to provide a service, including but not limited to
delivery of content, through electronic communication with a client
other than You.'

Apple's summary:

`"External Deployment" is defined to cover the external distribution of
APSL'ed code or use of APSL'ed code to provide a service (including
content delivery) to a third party through electronic communication with
that party.'

The FSF claims:

`In version 2.0 of the APSL, the definition of "Externally Deployed" has
been narrowed in a way that is appropriate for the respect of users'
freedoms. It has always been the position of FSF that the freedom of
Free Software is primarily for the users of that software. Technologies,
like web applications, are changing the way that users interact with
software. The APSL 2.0, like the Affero GPL, seeks to defend the freedom
of those who use software in these novel ways, without unduly hindering
the users' privacy nor freedom to use the software.'

These are strong words and I am surprised this has always been the
position of the FSF when I believe the position was more accurately
represented as "You should also have the freedom to make modifications
and use them privately in your own work or play, without even mentioning
that they exist.": http://www.gnu.org/philosophy/free-sw.html

In the APSL 2.0 when one "Externally Deploys" modified code one must
"make Source Code of all Your Externally Deployed Modifications either
available to those to whom You have Externally Deployed Your
Modifications, or publicly available. Source Code of Your Externally
Deployed Modifications must be released under the terms of set forth in
this License, including the license grants set forth in Section 3 below,
for as long as you Externally Deploy the Covered Code or twelve (12)
months from the date of initial External Deployment, whichever is
longer. You should preferably distribute the Source Code of Your
Externally Deployed Modifications electronically (e.g. download from a
web site);"

The Free Software Foundation's position is now clear: A licence can be a
Free Software licence even if it requires that non-distributed modified
code be supplied to an entity that is provided with an electronic
communication service. If the Google's of this world used code under
such a licence they would have to make their source code available to me
after I performed a search query and was delivered some content.

This is a hugely controversial issue. And one that shouldn't be decided
without vigorous debate.

What was a substantial freedom as part of GNU philosophy--"the freedom
to make modifications and use them privately in your own work or play,
without even mentioning that they exist"--is now only useful to hermits
and leeches. Anyone contributing by providing an electronic service
would no longer have any expectation of being able to keep modifications
private.

This "way that is appropriate for the respect of users' freedoms" and
this claim that "freedom of Free Software is primarily for the users of
that software" disguises decreasing concern for developer freedoms. Any
users of free software who modify their software are also developers and
they should be able to make some forms of modification without
distribution and without having to publish their code simply because
they provide a service through electronic communication. I don't know
where the line should be drawn but I am presently confident the Free
Software Foundation has crossed it.

I am not going to speculate about the motivations of the Free Software
Foundation because it simply opens me up to indefensible criticism.
Here's a mere consequence: If Debian is persuaded that the APSL 2.0 is
DFSG-free then a subsequent revision of the GPL with the addition of a
viral electronic service clause would also be DFSG-free.

Regards,
Adam

Lynn Winebarger

2003-08-07 10:23:15 UTC

Post by Adam Warner
What was a substantial freedom as part of GNU philosophy--"the freedom
to make modifications and use them privately in your own work or play,
without even mentioning that they exist"--is now only useful to hermits
and leeches. Anyone contributing by providing an electronic service
would no longer have any expectation of being able to keep modifications
private.

Of course, this all depends on what the arguer believes "private"
means. You don't need to attribute nefarious motives to rms - he (and
others) may simply disagree with you about when and how an activity/use
stops being private.
As for "hermits and leeches", why are casting aspersions? What do you
mean by "contributing by providing ..." if the modifications are not
released? Why isn't that being a leech? There are a lot of programs
that can do interesting things without having anything to do with a
network.

Lynn

Adam Warner

2003-08-07 12:19:54 UTC

Post by Lynn Winebarger

I specifically avoided speculating about Stallman's motives or even
singling him out from this overall decision by the Free Software
Foundation (apart from initially quoting Apple's thanks for his many
helpful comments in the process). I ended with a consequence. You are
free to speculate as to motives. Just please don't claim I'm the one
attributing "nefarious motives to rms."

Feel free to consider what "private" meant and now means. Feel free to
consider whether the Free Software Foundation made plain years ago that
a free software licence could enforce disclosure of source code if the
software as some part of a larger work performed in any way an
electronic communication service with another client.

[Note: client is undefined in the APSL. But the sentence where it
appears says "client other than You." Now You is "an individual or a
legal entity exercising rights". So the client is likely not another
computer program but another legal entity you provide a service to.]

Post by Lynn Winebarger
As for "hermits and leeches", why are casting aspersions?

It was a colourful flourish. Hermit is not a derogatory term (and
hermits will not be affected if they don't communicate). Leech was used
to indicate a scenario where users would en masse demand source code
merely because they used an electronic communication service.

Post by Lynn Winebarger
What do you mean by "contributing by providing ..." if the
modifications are not released? Why isn't that being a leech?

Do websites provide an electronic communication service? Do they
privately modify code? Why is that being a leech?

Post by Lynn Winebarger
There are a lot of programs that can do interesting things without
having anything to do with a network.

Conceded. But in a networked world how useful is a freedom that only
applies so long as you don't network with another person?

Regards,
Adam

Matthew Palmer

2003-08-07 13:18:27 UTC

Post by Adam Warner

Post by Lynn Winebarger
As for "hermits and leeches", why are casting aspersions?

Which isn't necessarily a bad thing. Remember how quickly the WWW got up
and became a very large part of Internet life? That growth has been
attributed in part to the fact that anyone could learn HTML by looking at
the source for pages produced by other people.

I don't think that such a licence term is particularly egregious. Look at
the GPL requirement - if you get the binary, you can get the source. Now,
who gets binaries? Users. So, users get the source to the programs they're
using. Now move to a webapp. Who uses/views the webapp? Users. So, under
the APSL 2.0, users get the source to the programs they're using.

Handwaving it may be, but I think the intent is similar.

Now, on the other side of the coin, the worry may be that it's important for
webapp vendors to be able to lock away their secrets because it's a
competitive market, or they think merely that they shouldn't have to give
the source away because of this reason or that reason. But they have
choices. They don't have to use GPL (or APSL 2.0) code in their product,
and/or they don't have to give every man and his dog access to it on the
Internet.

Post by Adam Warner

Post by Lynn Winebarger
There are a lot of programs that can do interesting things without
having anything to do with a network.

Conceded. But in a networked world how useful is a freedom that only
applies so long as you don't network with another person?

Not very. But I think you're reading the intent of the GPL the wrong way.
The FSF chant is "the users get the source". The GPL was written in a time
when the web didn't exist, and it was impossible to foresee this way of
distributing applications. I think this clause of the APSL 2.0 merely
brings GPL concepts into the modern era.

Post by Adam Warner
Regards,
Adam
--

Brian T. Sniffen

2003-08-07 15:10:34 UTC

Post by Matthew Palmer
I don't think that such a licence term is particularly egregious. Look at
the GPL requirement - if you get the binary, you can get the source. Now,
who gets binaries? Users. So, users get the source to the programs they're
using. Now move to a webapp. Who uses/views the webapp? Users. So, under
the APSL 2.0, users get the source to the programs they're using.
Handwaving it may be, but I think the intent is similar.

If your premises were true, I'd agree with your conclusion. But I've
never believed that renting someone an account on my system requires me
to provide source to all the GNU tools installed there. This is a
radical change. It affects not only the web, but encumbers every
other service, currently existing or not yet imagined, running over a
network.

For example, if this clause were in the GPL, it would require me to
provide full source to the Linux kernel, Emacs, gawk, Perl, etc.
Yuck! What a burden on providing access to a computer. In addition,
would this require source to mailing list software be distributed to
subscribers? How about spammers; they're using it, right?

If I set up a weather-monitoring device with an embedded Linux kernel,
and it publishes temperature and air pressure data over SNMP, do I
have to provide the source to those?

If I send a message like this one, which quotes yours, do I need to
send you the source for Emacs and Gnus? How about my MTA? If there
are routers in between which use APSL code, are you entitled to the
source for them too?

Post by Matthew Palmer
The FSF chant is "the users get the source". The GPL was written in a time
when the web didn't exist, and it was impossible to foresee this way of
distributing applications. I think this clause of the APSL 2.0 merely
brings GPL concepts into the modern era.

I think this era isn't very different from that of 15 years ago. RMS,
and the FSF, are spooked by the success of web service providers.
They didn't seem very upset by modems, remote terminals, and
timesharing systems, though. I think they're just experiencing
culture shock, and are overreacting to something which really isn't an
important change.

Worse, they're adding a sufficient encumbrance to networking computer
systems to lock code available only under an APSL/Affero style license
out of networked environments. If they succeed in promulgating these
ideas, they'll hinder growth of networked systems. Perhaps a good way
of summing up the problem is this:

They're discriminating against a field of endeavor. Now, it's Free to
discriminate against a business model, such as "A monopoly on software
in boxes on shelves." It's not Free to discriminate against a use
model, such as "running nuclear power plants." This is discrimination
against both a business model (web services providers) and a use model
(providing access to computers over a network).

-Brian

--
Brian T. Sniffen ***@alum.mit.edu
http://www.evenmere.org/~bts/

Adam Warner

2003-08-07 16:19:13 UTC

On Fri, 2003-08-08 at 03:10, Brian T. Sniffen wrote:
...

Post by Brian T. Sniffen
I think this era isn't very different from that of 15 years ago. RMS,
and the FSF, are spooked by the success of web service providers.
They didn't seem very upset by modems, remote terminals, and
timesharing systems, though. I think they're just experiencing
culture shock, and are overreacting to something which really isn't an
important change.
Worse, they're adding a sufficient encumbrance to networking computer
systems to lock code available only under an APSL/Affero style license
out of networked environments. If they succeed in promulgating these
ideas, they'll hinder growth of networked systems. Perhaps a good way
They're discriminating against a field of endeavor. Now, it's Free to
discriminate against a business model, such as "A monopoly on software
in boxes on shelves." It's not Free to discriminate against a use
model, such as "running nuclear power plants." This is discrimination
against both a business model (web services providers) and a use model
(providing access to computers over a network).

This is a brilliant summary. If your compelling reasoning is correct
you're not only placed the issue into the context of the DFSG but done
so in a straightforward way within one paragraph.

Wow!

Regards,
Adam

Branden Robinson

2003-08-07 20:36:00 UTC

Post by Adam Warner

Post by Brian T. Sniffen
They're discriminating against a field of endeavor. Now, it's Free to
discriminate against a business model, such as "A monopoly on software
in boxes on shelves." It's not Free to discriminate against a use
model, such as "running nuclear power plants." This is discrimination
against both a business model (web services providers) and a use model
(providing access to computers over a network).

This is a brilliant summary. If your compelling reasoning is correct
you're not only placed the issue into the context of the DFSG but done
so in a straightforward way within one paragraph.
Wow!

Hear, hear. <AOL>.

--
G. Branden Robinson | Men use thought only to justify
Debian GNU/Linux | their wrong doings, and speech only
***@debian.org | to conceal their thoughts.
http://people.debian.org/~branden/ | -- Voltaire

M. Drew Streib

2003-08-07 16:32:25 UTC

Post by Brian T. Sniffen
out of networked environments. If they succeed in promulgating these
ideas, they'll hinder growth of networked systems. Perhaps a good way

I could agree with you, except that networked systems can't really
be hindered too much now. They are pretty much a given.

Apple isn't so much discriminating against a use model, as discriminating
against _all_ use, in either a networked or distribution model, without
distributing source. Think of it as discriminating against the
business model of 'service', rather than the use of networked software.

They're simply cutting off the common GPL bypass these days, which
basically lets ASPs, web services, etc basically use GPL'd software
with no source releases (since no binaries are ever 'distributed' as
such). Since most of the net seems to be moving towards service models
and away from distribution models, this is merely a licence trying
to catch up.

-drew

--
M. Drew Streib <***@dtype.org>
Independent Rambler, Software/Standards/Freedom/Law -- http://dtype.org/

Mark Rafn

2003-08-07 17:05:22 UTC

Post by M. Drew Streib

Post by Brian T. Sniffen
out of networked environments. If they succeed in promulgating these
ideas, they'll hinder growth of networked systems. Perhaps a good way

I could agree with you, except that networked systems can't really
be hindered too much now. They are pretty much a given.

Indeed. These ideas will severly hinder growth of software which is
licensed under such terms.

If Debian starts accepting such terms, it removes one of the primary
reasons I choose it over other distributions (leaving the other reason,
which is that it works well). Once I have to start examining the license
of every piece of software I run on a server, I may as well just buy a
proprietary system. At least there I know I won't be forced to distribute
anything.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Brian T. Sniffen

2003-08-07 18:05:01 UTC

Post by M. Drew Streib

Post by Brian T. Sniffen
out of networked environments. If they succeed in promulgating these
ideas, they'll hinder growth of networked systems. Perhaps a good way

I could agree with you, except that networked systems can't really
be hindered too much now. They are pretty much a given.

In that case, won't it hinder the use of APSL/Affero GPL licensed
software in networked environments?

Post by M. Drew Streib
Apple isn't so much discriminating against a use model, as discriminating
against _all_ use, in either a networked or distribution model, without
distributing source. Think of it as discriminating against the
business model of 'service', rather than the use of networked software.
They're simply cutting off the common GPL bypass these days, which
basically lets ASPs, web services, etc basically use GPL'd software
with no source releases (since no binaries are ever 'distributed' as
such). Since most of the net seems to be moving towards service models
and away from distribution models, this is merely a licence trying
to catch up.

So why hinder a typesetter who returns his work as PDF more than a
typesetter who returns printed pages? Why favor an HTML file
distributed on a floppy over one distributed via HTTP? This
insistence that interacting with software over a network of electrons
is somehow different from interacting with software via DHL is
ridiculous. It's not a license catching up, or closing loopholes
without impacting freedom: it's that license authors saw something
which bothered them, and are prohibiting it in their licenses.
They're allowed to do that, certainly, but that doesn't make it Free.

-Brian

--
Brian T. Sniffen ***@alum.mit.edu
http://www.evenmere.org/~bts/

MJ Ray

2003-08-07 10:51:33 UTC

Post by Adam Warner
Here's a mere consequence: If Debian is persuaded that the APSL 2.0 is
DFSG-free then a subsequent revision of the GPL with the addition of a
viral electronic service clause would also be DFSG-free.

It is expected that GPL-3 will contain something similar to the Affero GPL
requirement for remote services to offer users the code. Do you object
to that? If so, why? If you are offering interaction with the code
via some sort of remote procedure call, you are not using it privately
for your own ends and some of the users may want to adapt the software,
which is a freedom normally offered by free software.

I think you are unwise to throw words like "leeches" and "viral" around
so freely. That sort of thing spoiled an otherwise interesting email
for me.

--
MJR/slef My Opinion Only and possibly not of any group I know.
http://mjr.towers.org.uk/ jabber://***@jabber.at

Adam Warner

2003-08-07 13:16:21 UTC

I am not prepared to answer these questions at this time. If I had to
make a snap decision it would be for the status quo that licensing
obligations apply upon source code distribution.

The issue at hand is the superset that the Free Software Foundation has
now declared to be a Free Software licence:

"(b) to use Covered Code, alone or as part of a Larger Work, in any way
to provide a service, including but not limited to delivery of content,
through electronic communication with a client other than You."

As I noted in my previous message, client appears to be a different
individual or legal entity, not a different computer program.

This appears to be a very broad ("in any way") test for a requirement to
provide source code when electronically communicating with a client ("in
any way to provide a service, including but not limited to delivery of
content"). As I wrote, "If the Google's of this world used code under
such a licence they would have to make their source code available to me
after I performed a search query and was delivered some content." This
appears to be a reasonable application of the paragraph even though it
would have extraordinary consequences.

It's up to Debian developers to decide. I've finished highlighting what
I consider important. I wasn't even going to raise the issue until Jens
Schmalzing forced it by inquiring about packaging some APSL software [of
course there's always non-free, and I agree with the Debian Social
Contract that "although non-free software isn't a part of Debian, we
support its use, and we provide infrastructure (such as our bug-tracking
system and mailing lists) for non-free software packages." I don't
believe there is stigma attached to packaging software classified as
non-free and in many cases it is just as accessible to Debian users.]

Post by MJ Ray
I think you are unwise to throw words like "leeches" and "viral" around
so freely. That sort of thing spoiled an otherwise interesting email
for me.

That's a pity.

Regards,
Adam

Matthew Palmer

2003-08-07 13:28:18 UTC

Post by Adam Warner
The issue at hand is the superset that the Free Software Foundation has
"(b) to use Covered Code, alone or as part of a Larger Work, in any way
to provide a service, including but not limited to delivery of content,
through electronic communication with a client other than You."
As I noted in my previous message, client appears to be a different
individual or legal entity, not a different computer program.

Taken to an extreme, this paragraph could be interpreted to mean that, if I
use APSL'd source code to track client contacts internally to my business,
then e-mail a client, the code for my contact tracking could be requested by
the client. Whether that's a particularly reasonable interpretation or not
would be up to the legal beagles. I don't think it would really pass
muster, since the link is so tenuous, but you never know.

And then you can argue how much of the communication needs to be electronic?
Does a phone count, since it uses electrons? If I write a letter on my
computer, print it, and then mail it, is that still electronic
communication? If you argue no, because it involves a human element, note
that I used my fingers to type the e-mail, and someone had to click their
mouse to retrieve the content, so that's a human element too...

Boy licence debating is fun!

Post by Adam Warner
content"). As I wrote, "If the Google's of this world used code under
such a licence they would have to make their source code available to me
after I performed a search query and was delivered some content." This
appears to be a reasonable application of the paragraph even though it
would have extraordinary consequences.

Would you be able to detail these extraordinary (I presume you mean
negative) consequences?

In my view, that Google example you give is exactly the sort of thing that
the APSL is aiming for - if you provide a service using the APSL'd code, you
have to provide that APSL'd code to any user of that service who requests
it.

As regards DFSG-freeness, I don't think there's anything in the guidelines
which is explicitly opposed to this. It touches a fair few edges, and I can
certainly understand how it could push some people's buttons, but my vote
would be close, but not quite non-free. Other clauses in the licence may
be nasty, though - I haven't read the whole licence.

- Matt

MJ Ray

2003-08-07 14:19:07 UTC

Post by Adam Warner
I am not prepared to answer these questions at this time. If I had to
make a snap decision it would be for the status quo that licensing
obligations apply upon source code distribution.

I'm puzzled by this phrasing. Don't you mean binary (or any)
distribution? At that point, many licences put obligations upon you.

Post by Adam Warner
The issue at hand is the superset that the Free Software Foundation has

Can you demonstrate where this contradicts their Free Software Definition?
This does not appear to be news, so I am surprised that you feel a need
to highlight it on this list.

Post by Adam Warner
This appears to be a very broad ("in any way") test for a requirement to
provide source code when electronically communicating with a client ("in
any way to provide a service, including but not limited to delivery of
content").

I agree that the wording of this part seems a little broad. It is more
a question of whether we can take it at face value. The Google problem
is well-known, I thought? There are all these millions of users who are
not able to adapt the software they use...

[...]

Post by Adam Warner
believe there is stigma attached to packaging software classified as
non-free and in many cases it is just as accessible to Debian users.]

I disagree with you. I think packaging for non-free does little to help
us achieve our aims. I would love to see non-free go away, but that's
a different discussion.

--
MJR/slef My Opinion Only and possibly not of any group I know.
http://mjr.towers.org.uk/ jabber://***@jabber.at

Adam Warner

2003-08-07 15:29:02 UTC

I'm puzzled by this phrasing. Don't you mean binary (or any)
distribution? At that point, many licences put obligations upon you.

Thanks for picking this up. My brain wasn't engaged when I wrote that
(and it's even less engaged at this time of the morning). I meant the
standard types of distribution (binary and source) as distinct from
deployment. For example the GPL imposes no obligation in the absence of
distribution of a derivate work.

Post by Adam Warner
The issue at hand is the superset that the Free Software Foundation has

Can you demonstrate where this contradicts their Free Software Definition?
This does not appear to be news, so I am surprised that you feel a need
to highlight it on this list.

It's news. Earlier versions of the APSL were declared non-free because
"The APSL does not allow you to make a modified version and use it for
your own private purposes, without publishing your changes.":
http://www.fsf.org/philosophy/historical-apsl.html

You may have also missed the paragraph in the Free Software Definition
which states: "You should also have the freedom to make modifications
and use them privately in your own work or play, without even mentioning
that they exist. If you do publish your changes, you should not be
required to notify anyone in particular, or in any particular way."
http://www.gnu.org/philosophy/free-sw.html

Distributing nothing while simply providing an electronic communications
service may not be related to publishing changes.

As I said to Lynn, feel free to consider what "private" meant and now
means. Feel free to consider whether the Free Software Foundation made
plain years ago that a free software licence could enforce disclosure of
source code if the software as some part of a larger work performed in
any way an electronic communication service with another client.

It's clear to me that a convenient redefinition has occurred. I'm not
debating it with you.

Yes, and it's OK. I'd rather have that than a leech mentality that
promotes entitlement to source code simply because one accesses an
electronic communications service.

Post by MJ Ray
[...]

Post by Adam Warner
believe there is stigma attached to packaging software classified as
non-free and in many cases it is just as accessible to Debian users.]

I disagree with you. I think packaging for non-free does little to help
us achieve our aims. I would love to see non-free go away, but that's
a different discussion.

That you had to snip the Debian Social Contract in order to disagree
with me is telling.

And by the way that's "I don't believe..."

I'm content to disagree with you. I'll reconsider after Debian's Social
Contract is rewritten.

Regards,
Adam

MJ Ray

2003-08-07 16:16:41 UTC

Post by Adam Warner
It's news. Earlier versions of the APSL were declared non-free because
"The APSL does not allow you to make a modified version and use it for

It's not news. It may have been news when Affero GPL was listed, but
I don't see anything changed now.

[...]

Post by Adam Warner
As I said to Lynn, feel free to consider what "private" meant and now
means. Feel free to consider whether the Free Software Foundation made
plain years ago that a free software licence could [...]

I'll bow to your superior knowledge of past FSF pronouncements. When do
you feel that they were unclear that their aim was for users to have
access to the source code of the software that they used?

Maybe some of them erred in the past, when the web was young. If so,
then this correction is a good thing, isn't it? They're not forever
bound by past mistakes which are in conflict with their aims.

[Google problem]

Post by Adam Warner
Yes, and it's OK. I'd rather have that than a leech mentality that
promotes entitlement to source code simply because one accesses an
electronic communications service.

Funny. Most people talk about "using google" instead of "accessing
the google site" or similar. I'd rather we didn't slip back to begging
software lords for changes.

[...non-free...]

Post by Adam Warner
That you had to snip the Debian Social Contract in order to disagree
with me is telling.

Only of your excessive word count. Sorry for snipping in a bad place.
The references header was intact. Anyway, I know what the SC says and
those are the bounds we work in. Doesn't mean that I have to do anything
for it, like it or encourage it.

--
MJR/slef My Opinion Only and possibly not of any group I know.
http://mjr.towers.org.uk/ jabber://***@jabber.at
Creative copyleft computing services via http://www.ttllp.co.uk/
Thought: Edwin A Abbott wrote about trouble with Windows in 1884

Stephen Ryan

2003-08-08 01:37:20 UTC

I have a web site on a server that I control. Every piece of software
on that server is there to assist in rendering or managing the web site
in some way or another. I'm aware that Apache and the Linux kernel are
not under an Affero-type license, but suppose with me for a moment that
the whole installation was under such a license. I'm now liable to
distribute the source code for an entire operating system to every
person who manages to obtain a web page from me. I'm also liable to
distribute the source code for an entire operating system to every
sco'er in the world who manages to send me a spam, because they're
"using" my email service. If such a bit is added to the GPL with no
other changes, I either have to deliver the source code with every web
page (or spam), or I have to promise to keep it up there for a minimum
of three years. All of a sudden, licensing SCO or Windows looks cheap
by comparison - at least they only demand the large payment once.

Thankfully, this is all hypothetical at this point, and the kernel and
Apache are both licensed under a different license (GPLv2 only, and
Apache has its own license), but I'm bothered by the fact that this is
being put forth as "free". As Adam points out, this is a networked
world; if I can only afford to exercise my freedom by not being
networked (being a hermit), that freedom is *worthless*.

MJ Ray

2003-08-07 14:22:40 UTC

Post by Stephen Ryan
the whole installation was under such a license. I'm now liable to
distribute the source code for an entire operating system to every
person who manages to obtain a web page from me.

How does this differ from your current obligation to either provide
the source or equivalent offer to that which you obtained the source?
Or is your server a 0-user affair? So why wouldn't the offer clause
work for you?

--
MJR/slef My Opinion Only and possibly not of any group I know.

Brian T. Sniffen

2003-08-07 15:28:41 UTC

How does this differ from your current obligation to either provide
the source or equivalent offer to that which you obtained the source?

Well, for a start, the Affero/APSL clause is a restriction on use: the
GPL grants only freedoms from the restrictions imposed by Copyright
law. The Affero and APSL license impose additional restrictions in
exchange for granting some freedoms.

Post by MJ Ray
Or is your server a 0-user affair? So why wouldn't the offer clause
work for you?

The APSL doesn't have an offer clause. And even if it did, hosting an
entire source distribution with working build scripts because I want
to have a listserv is an unreasonable burden.

-Brian

--
Brian T. Sniffen ***@alum.mit.edu
http://www.evenmere.org/~bts/

Stephen Ryan

2003-08-07 16:19:58 UTC

Has Debian made such an offer? I can only pass along such an offer if I
received one in the first place. To my knowledge, Debian has made no
such offer on any code. It also does not apply if I make any commercial
use of my server (which I do, to the tune of a couple hundred dollars
per year).

The difference is that I don't currently run a debian mirror on it, and
I don't have to, either. The difference is that I currently have to get
into the distribution business only if I want to be in the distribution
business, whereas such licenses obligate every person who has a computer
running such code to be in the distribution business (I'm phrasing this
carefully so as to avoid any difficulty over the definition of a "user"
here).

Of what use is "Free" software if nobody is willing to run it?

Mark Rafn

2003-08-07 16:34:01 UTC

I haven't read the license fully yet, so please ignore me if this issue is
addressed, or there's a phrasing that makes it unimportant.

What, exactly, do you have to distribute to "users" under this clause?
Does it not have to be the complete service itself? How does that exclude
the data and configuration information (e.g. passwords, customer
information, etc.) that are necessary to actually run the service?
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-07 14:55:56 UTC

It is expected that GPL-3 will contain something similar to the
Affero GPL requirement for remote services to offer users the code.
Do you object to that? If so, why?

For what it's worth, I think the Affero bit has problems that the APSL
2.0 does not -- namely that the Affero bit places a restriction on the
code itself that if quine-like functionality exists it must be
preserved. From that perspective, the APSL 2.0 is an improvement. On
the other hand, the APSL doesn't seem to address the very problem the
quine-bit solves: that the copyright holder gets to decide on a
case-by-case basis whether it goes in. I'm not sure that that really
makes much difference though, since the copyright holder also gets to
choose the license.

I personally think that this move (APSL 2.0 & Affero both) is a stupid
one. I think it will cause greater problems using old code in new
projects (due to license incompatibility and uncertainty) and in the
end probably wont be necessary. But I'd like to allow events to
decide themselves. So I'm willing, in principle, to consider
something like this free, though I don't know if other folks here
agree with me on this.

But I am curious what Apple thinks would happen if you used APSL 2.0
covered code in a web server (let's assume it's just a basic web
server). Would that qualify as offering a service? "Externally
Deploy" is defined in terms of distribution or something that will "in
any way provide a service". Is the phrase "provide a service" clear
and well defined enough that it would include web apps (e.g., Google,
which I do think they mean to include) but not web servers or inetd?
That's how I interpret it, though there are some fuzzy cases (e.g., a
mail-order typesetting business) which I'm happy leaving fuzzy.

If Apple intends to incorporate patches under this license, one hopes
that they wont want the definition of "provide a service" to be too
extensive either. I.e., if the definition of "provide a service"
evolves it will hopefully be a reasonable evolution. So I guess in
the end I'm cautiously positive, though I'm curious what others think.
I know some folks here were very much against the Affero GPL, and I
don't know if this answers their problems with it.

Note: I haven't looked over the rest of the APSL either. Also, IANAL,
IANADD. ;)

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Bernhard R. Link

2003-08-07 15:20:46 UTC

Post by MJ Ray
It is expected that GPL-3 will contain something similar to the Affero GPL
requirement for remote services to offer users the code.

Hopefully not. (When I remember the "PHPNuke licese"-thread there
was some word that it will not.)

Post by MJ Ray
Do you object to that? If so, why?

At least I do, because I as user (in the only undoubtly sense of
user as those who owns the copy and runs it) am loosing elementary
rights. (For example I might no longer be able to run it at all,
as I cannot fullfill the requirements).

It's the idea of copyleft to give the (in some opinion unnaturaly
restricted by law) right to make copies and distribute them in
exchange for the limitations to give source together with binaries.

While I can live with this restriction as I can respect the
(in some eyes stupid) limitation on make copies of other people's
work, putting restrictions on the usage of a legal copy is simply
immoral. (And while the lawyers seem currently struggling here in
Germany if usage of a program is copying to RAM and thus limited
by copyright law or not, I just handle immoral things as void).

Post by MJ Ray
If you are offering interaction with the code
via some sort of remote procedure call, you are not using it privately
for your own ends and some of the users may want to adapt the software,
which is a freedom normally offered by free software.

It is only offered for those having copies. And I'd be sadly supprised
if for example giving students accounts in a computer lab would
count as distribution. (And remembering what I saw in some /bin/clear
scripts I doubt any jurisdiction will).

Hochachtungsvoll,
Bernhard R. Link

--
Sendmail is like emacs: A nice operating system, but missing
an editor and a MTA.

Anthony DeRobertis

2003-08-07 20:01:11 UTC

Post by Bernhard R. Link
(And while the lawyers seem currently struggling here in
Germany if usage of a program is copying to RAM and thus limited
by copyright law or not, I just handle immoral things as void).

Really? Just now? In the US, Title 17 says it's not. At least I'm
pretty sure it does, not online at the moment to check.

Arnoud Galactus Engelfriet

2003-08-08 08:03:31 UTC

Post by Anthony DeRobertis

Really? Just now? In the US, Title 17 says it's not. At least I'm
pretty sure it does, not online at the moment to check.

17 USC 117 says it is not an infringement to make copies if they
are "created as an essential step in the utilization of the
computer program in conjunction with a machine". I would argue
this covers any loading from harddisk into RAM, so the copyright
holder cannot forbid you to do this.

European copyright law with respect to computer programs is
harmonized based on a.o. Directive 91/250/EEC of 14 May 1991.
Article 4 of this Directive says quite explicitly:

Insofar as loading, displaying, running, transmision or storage
of the computer program necessitate such reproduction, such
acts shall be subject to authorization by the rightholder.

Article 5 then says that the authorization is not necessary if
a lawful acquirer needs to perform the acts for the use of
the program in accordance with its intended purpose.

So, if "running" a program necessitates "reproduction" in
the meaning of copyright law, you need permission from the
copyright holder, unless you have lawfully acquired the
program and you're using it as intended.

Arnoud

--
Arnoud Engelfriet, Dutch patent attorney - Speaking only for myself
Patents, copyright and IPR explained for techies: http://www.iusmentis.com/

Mark Rafn

2003-08-07 16:19:54 UTC

It is expected that GPL-3 will contain something similar to the Affero GPL
requirement for remote services to offer users the code.

Indeed, and this is the primary reason I recommend authors NOT add the "or
any later version" when choosing GPLv2.

Post by MJ Ray
Do you object to that? If so, why?

Vehemently. It removes the ability of users to privately modify work,
which IMO is simply not free. Almost any piece of software in a business
is used (indirectly in many cases, but used nonetheless) by most of it's
customers.

http://lists.debian.org/debian-legal/2003/debian-legal-200303/msg00805.html
is a list of software "uses" that are hard to distinguish from each
other in a license, so would all require full source to be made publicly
available.

Post by MJ Ray
I think you are unwise to throw words like "leeches" and "viral" around
so freely. That sort of thing spoiled an otherwise interesting email
for me.

Is there any possible reason to prefer this requirement other than to
punish "leeches"? This seems a hugely bad motivation for a major license
change.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-07 17:06:43 UTC

Post by MJ Ray
Do you object to that? If so, why?

Vehemently. It removes the ability of users to privately modify work,
which IMO is simply not free. Almost any piece of software in a business
is used (indirectly in many cases, but used nonetheless) by most of it's
customers.
http://lists.debian.org/debian-legal/2003/debian-legal-200303/msg00805.html
is a list of software "uses" that are hard to distinguish from each
other in a license, so would all require full source to be made publicly
available.

What are you trying to say here?

* That providing a service in this context necessarily includes the
mail-order typesetting scenario?

* That what "providing a service" means here isn't really nailed down,
and reasonable people might include the mail-order typesetting
scenario?

* That even though reasonable people would disagree, we can't trust
Apple (or other licensors) not to include the mail-order typesetting
scenario in "providing a service"?

If it's the first, I think you're being silly. If it's the second,
I'm sceptical, but willing to listen to more argument. If it's the
third, I'm worried about this myself, but cautiously willing to give
folks the benefit of the doubt.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Mark Rafn

2003-08-07 19:09:05 UTC

Post by Mark Rafn
http://lists.debian.org/debian-legal/2003/debian-legal-200303/msg00805.html
is a list of software "uses" that are hard to distinguish from each
other in a license, so would all require full source to be made publicly
available.

What are you trying to say here?
* That providing a service in this context necessarily includes the
mail-order typesetting scenario?

Of course it does. Why would delivery via paper confer fewer rights on
the user than delivery by email or HTTP?

Post by Jeremy Hankins
* That what "providing a service" means here isn't really nailed down,
and reasonable people might include the mail-order typesetting
scenario?

Certainly true.

Post by Jeremy Hankins
* That even though reasonable people would disagree, we can't trust
Apple (or other licensors) not to include the mail-order typesetting
scenario in "providing a service"?

Also true, but I think it's more about the fundamental problem that this
is a non-free restriction than about abuse by licensors.

Post by Jeremy Hankins
If it's the first, I think you're being silly.

Perhaps, but I'm not laughing. I honestly don't see why you'd expect, for
example, someone who gets a statement electronically to have more access
to a billing system than someone who gets it via snail-mail.

Post by Jeremy Hankins
If it's the second,
I'm sceptical, but willing to listen to more argument.

I believe myself to be reasonable, and I don't see any fundamental
difference between delivering printeed page of output and viewing a
webpage of output. Both are use of software, and neither should require
distribution of the software so used.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-07 20:08:20 UTC

What are you trying to say here?
* That providing a service in this context necessarily includes the
mail-order typesetting scenario?

Of course it does. Why would delivery via paper confer fewer rights on
the user than delivery by email or HTTP?

Well, the APSL specifically says that the service must be "through
electronic communication" to qualify:

http://www.opensource.apple.com/apsl/2.0.txt

Though that was as much my mistake as yours, for choosing my example
carelessly. How about a web server, instead? Do you think that
using a web server to make your content available to others qualifies
as providing a service? Do you think Apple thinks so?

In the list you referenced, the service goes electronic when Joe
receives the document via email, munges it, and sends it back. Even
there, I think it's hard to claim that Joe is using the "Covered Code,
alone or as part of a Larger Work, in any way to provide a service."
It's only when Joe sets up a procmail recipe that automatically
munges, and then sends back the results, that the APSL is triggered.
IMHO, at any rate.

Post by Mark Rafn
Also true, but I think it's more about the fundamental problem that this
is a non-free restriction than about abuse by licensors.

I'm not convinced we can clearly get non-free out of the DFSG on this
one. I don't buy the discrimination against fields of endeavor, and
unlike the affero GPL this isn't a restriction on modification.

It's a restriction, yes. And not one I particularly like, if the
truth be known. But the analogy between this restriction and the
source-redistribution restriction of the GPL is simply too strong for
me to ignore it. If you assume that the definition of "Externally
Deploy" (or more specifically, "provide a service") is going to be
reasonable I have trouble seeing where you can say it's not DFSG free.

I'm ready to say that the license is a bit ambiguous and needs
clarification, but I'm not convinced that what Apple is trying to do
is non-free.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Brian T. Sniffen

2003-08-07 21:16:16 UTC

Post by Jeremy Hankins
How about a web server, instead? Do you think that
using a web server to make your content available to others qualifies
as providing a service? Do you think Apple thinks so?
In the list you referenced, the service goes electronic when Joe
receives the document via email, munges it, and sends it back.

So a hypothetical Amazon 1990, which receives a request over e-mail
and responds by sending a package via physical mail, is not an
electronic service? Neither is Gutenberg-USPS, which will e-mail me a
document in response to a physical request, no SASE required?

Post by Jeremy Hankins
Even there, I think it's hard to claim that Joe is using the
"Covered Code, alone or as part of a Larger Work, in any way to
provide a service."

This confuses me. How can you not say, when Joe's using the covered
code to perform typesetting for others, that he's not using it in any
way to provide a service?

Post by Mark Rafn
Also true, but I think it's more about the fundamental problem that this
is a non-free restriction than about abuse by licensors.

I'm not convinced we can clearly get non-free out of the DFSG on this
one. I don't buy the discrimination against fields of endeavor, and
unlike the affero GPL this isn't a restriction on modification.
It's a restriction, yes. And not one I particularly like, if the
truth be known. But the analogy between this restriction and the
source-redistribution restriction of the GPL is simply too strong for
me to ignore it. If you assume that the definition of "Externally
Deploy" (or more specifically, "provide a service") is going to be
reasonable I have trouble seeing where you can say it's not DFSG free.

Copyright law does not grant any control over a third party's use, but
only on modification and distribution. The GNU GPL's
source-redistribution requirement only kicks in when attempting to do
something normally forbidden by copyright law. That is, it lifts the
barrier of copyright law, but only part way.

This license, the APSL, imposes a barrier on things copyright law
doesn't cover. What happens if I choose to refuse the license -- can
I ignore the APSL then? For example, let's say I hire Modifiers,
Inc. to take an APSL2-covered work and modify it. They comply with
the APSL, and post it on their web site. They also hand me a hard
drive with the software on it. I use that drive in a computer which
provides a web service. I decline to accept Apple's license to modify
or distribute their code. All I'm doing is using it, so they can't
touch me.

The above paragraph mostly says that the APSL is a bad idea and may be
unenforceable; if you don't buy that, at least consider the original
argument: that a restriction in addition to those imposed by copyright
law is necessarily non-free.

-Brian

--
Brian T. Sniffen ***@alum.mit.edu
http://www.evenmere.org/~bts/

Jeremy Hankins

2003-08-08 01:12:51 UTC

Given that I think the license is vague & needs clarification, I'll
assume you're asking what I think should be the case, not what the
license says. Because, as I said, the license is ambiguous. As I see
it we can either trust Apple[1], try to get clarification, or give
up on the license.

So no, neither of those should qualify as external deployment.

[1] As I said before, this may not be as unreasonable as it sounds
since Apple will possibly be on the receiving end of the license as
well.

Post by Jeremy Hankins
Even there, I think it's hard to claim that Joe is using the
"Covered Code, alone or as part of a Larger Work, in any way to
provide a service."

This confuses me. How can you not say, when Joe's using the covered
code to perform typesetting for others, that he's not using it in any
way to provide a service?

As I see it, yes, there is a difference. In one case it's automatic,
and the typesetting code is itself providing a service -- i.e., it's
directly being used by the customer. In the other Joe is interacting
with the typesetting code, not the customer, so Joe is providing the
service.

Post by Jeremy Hankins
It's a restriction, yes. And not one I particularly like, if the
truth be known. But the analogy between this restriction and the
source-redistribution restriction of the GPL is simply too strong for
me to ignore it. If you assume that the definition of "Externally
Deploy" (or more specifically, "provide a service") is going to be
reasonable I have trouble seeing where you can say it's not DFSG free.

The APSL refers several times to performance, which suggests that they
intend a public-performance argument to use that to back up this
requirement.

Post by Brian T. Sniffen
The above paragraph mostly says that the APSL is a bad idea and may be
unenforceable; if you don't buy that, at least consider the original
argument: that a restriction in addition to those imposed by copyright
law is necessarily non-free.

Why?

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Brian T. Sniffen

2003-08-08 18:07:51 UTC

Post by Jeremy Hankins
Even there, I think it's hard to claim that Joe is using the
"Covered Code, alone or as part of a Larger Work, in any way to
provide a service."

This confuses me. How can you not say, when Joe's using the covered
code to perform typesetting for others, that he's not using it in any
way to provide a service?

That's an interesting idea, but it is not what is written there: the
APSL talks about using the software in any way to provide a service.
So when considering the question "Is Joe using the software in any way
to provide a service?", is "No" an answer you find reasonable? Can
you truly state "Joe is not using the software in any way to provide a
service"?

The APSL refers several times to performance, which suggests that they
intend a public-performance argument to use that to back up this
requirement.

That's an interesting idea. It's not obvious to me how to interpret
it; I'll have to think about it some more.

Why?

It's a convenient test, seems intuitively reasonable, and puts the GNU
GPL2, BSD, Artistic, etc. licenses on one side, and all those which
distribute against fields of endeavor on the other.

-Brian

--
Brian T. Sniffen ***@alum.mit.edu
http://www.evenmere.org/~bts/

Jeremy Hankins

2003-08-12 00:49:17 UTC

Post by Brian T. Sniffen
That's an interesting idea, but it is not what is written there: the
APSL talks about using the software in any way to provide a service.
So when considering the question "Is Joe using the software in any
way to provide a service?", is "No" an answer you find reasonable?
Can you truly state "Joe is not using the software in any way to
provide a service"?

All I'm trying to say is that there is a reasonable interpretation of
the license. What's more, I suspect that the stuff Apple cares about
falls under the reasonable interpretation. After all, if I understand
your objections, they'd be satisfied if Apple agreed that "providing a
service" doesn't apply to situations where a person is using the
software to manually provide a service (e.g., your email-based
typesetting service). Is that correct?

Am I trying to say that the license, as written, is clear & ok?
Absolutely not. But I do think that, given the basic structure of the
license and a bit of work ironing out details, it could be made DFSG
free. Personally, in addition to the above I'd like to see an
exemption for works that implement well-known services, but that would
be tricky and I don't think it's necessary for the DFSG.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Branden Robinson

2003-08-11 19:59:17 UTC

On Thu, Aug 07, 2003 at 05:16:16PM -0400, Brian T. Sniffen wrote:
[...]

if you don't buy that, at least consider the original argument: that a
restriction in addition to those imposed by copyright law is
necessarily non-free.

We don't uphold this principle in practice. DFSG #4 explicitly
allows the sort of "poor man's trademark" restriction that is exercised
by the Apache license.

At one point I considered writing a critique of the Apache license for
this reason, but figured it would just provoke flameage and not much in
the way of serious consideration.

Besides, a statement like "any restriction beyond that imposed by
copyright law is non-free" necessitatates that we answer the question
"which copyright law?".

I think we're better off with the
broad-principles-combined-with-specific-tests approach that I believe is
(imperfectly) exemplified by our current Social Contract and DFSG.

--
G. Branden Robinson |
Debian GNU/Linux | "Bother," said Pooh, as he was
***@debian.org | assimilated by the Borg.
http://people.debian.org/~branden/ |

Andrew Suffield

2003-08-11 20:24:21 UTC

Post by Branden Robinson
Besides, a statement like "any restriction beyond that imposed by
copyright law is non-free" necessitatates that we answer the question
"which copyright law?".

Plus it is fundamentally flawed, because it condones anything
forbidden by copyright law as being an acceptable restriction - when
it probably isn't, especially if you happen to live in the US.

--
.''`. ** Debian GNU/Linux ** | Andrew Suffield
: :' : http://www.debian.org/ |
`. `' |
`- -><- |

Mark Rafn

2003-08-07 20:56:06 UTC

Post by Jeremy Hankins
What are you trying to say here?
* That providing a service in this context necessarily includes the
mail-order typesetting scenario?

Of course it does. Why would delivery via paper confer fewer rights on
the user than delivery by email or HTTP?

Well, the APSL specifically says that the service must be "through

Ok, though this is an arbitrary distinction, and I'd argue that something
that restricts e-mail communication is no more free than something that
restricts snail-mail communication.

Anyway, rephrase my question to "why would delivery over e-mail confer
fewer rights to a user than delivery over http or RMI (or RMI over HTTP)?"

Post by Jeremy Hankins
Though that was as much my mistake as yours, for choosing my example
carelessly. How about a web server, instead? Do you think that
using a web server to make your content available to others qualifies
as providing a service?

Very much so.

Post by Jeremy Hankins
Do you think Apple thinks so?

I'd be shocked if they didn't, though I can't speak for them.

Post by Jeremy Hankins
In the list you referenced, the service goes electronic when Joe
receives the document via email, munges it, and sends it back.

How about if he delivers it by hand, and recieves it via e-mail?

Post by Jeremy Hankins
Even there, I think it's hard to claim that Joe is using the "Covered
Code, alone or as part of a Larger Work, in any way to provide a
service."

Really? I think it's hard to claim otherwise.

Post by Jeremy Hankins
It's only when Joe sets up a procmail recipe that
automatically munges, and then sends back the results, that the APSL is
triggered. IMHO, at any rate.

Interesting opinion, but it seems to me it's one based on what you'd like
to happen, rather than what the license actually says.

Post by Mark Rafn
Also true, but I think it's more about the fundamental problem that this
is a non-free restriction than about abuse by licensors.

It's a restriction on use (per definition 1.4 section b). DFSG has no
explicit item that use of the software must not be restricted, but any use
restriction completely breaks users' trust of the freeness of Debian.

Completely different. GPL is about distribution, and specifically says
that no use of the software is restricted. APSL limits use.

Post by Jeremy Hankins
If you assume that the definition of "Externally
Deploy" (or more specifically, "provide a service") is going to be
reasonable I have trouble seeing where you can say it's not DFSG free.

I haven't thought of any definition of "externally deploy" that is
reasonable. I suspect none exists.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-08 01:28:53 UTC

Post by Jeremy Hankins
Well, the APSL specifically says that the service must be "through

Ok, though this is an arbitrary distinction, and I'd argue that something
that restricts e-mail communication is no more free than something that
restricts snail-mail communication.

Email isn't entirely electronic unless it's also automatic. If you
type in the message and send it, there's a decidedly non-electronic
(well, non-digital) element: you.

Post by Mark Rafn
Anyway, rephrase my question to "why would delivery over e-mail confer
fewer rights to a user than delivery over http or RMI (or RMI over HTTP)?"

If it's via an automatic responder with no human intervention? No
reason at all.

Very much so.

Post by Jeremy Hankins
Do you think Apple thinks so?

I'd be shocked if they didn't, though I can't speak for them.

Well, who knows, you may be right. But I very much doubt that's
really the case they're after, even if they would think it's caught by
the license. A point worth clarifying.

Post by Jeremy Hankins
In the list you referenced, the service goes electronic when Joe
receives the document via email, munges it, and sends it back.

How about if he delivers it by hand, and recieves it via e-mail?

Nope.

Post by Jeremy Hankins
It's only when Joe sets up a procmail recipe that
automatically munges, and then sends back the results, that the
APSL is triggered. IMHO, at any rate.

Interesting opinion, but it seems to me it's one based on what you'd
like to happen, rather than what the license actually says.

It's also based on what I think Apple is most concerned about. But
yes, the license is certainly not clear on this point.

Post by Jeremy Hankins
I'm not convinced we can clearly get non-free out of the DFSG on
this one. I don't buy the discrimination against fields of
endeavor, and unlike the affero GPL this isn't a restriction on
modification.

It's a restriction on use (per definition 1.4 section b). DFSG has
no explicit item that use of the software must not be restricted,
but any use restriction completely breaks users' trust of the
freeness of Debian.

This seems like a very difficult argument to know when to make. Do
you apply it to any new license, suggesting that any license not
already in Debian will break users' trust of Debian's freeness?

Completely different. GPL is about distribution, and specifically says
that no use of the software is restricted. APSL limits use.

So? This is a serious question: why does that matter?

Note that there are two types of restrictions (either for
modifications or for use): restrictions *on* the modification or use,
and hurdles that must be jumped iff the work is modified/used.

The GPL places both kinds of restrictions on modification, but only
the latter is being placed on use by the APSL.

I haven't thought of any definition of "externally deploy" that is
reasonable. I suspect none exists.

Ok. Say that it must be electronic and automatic (i.e., no human step
in the process). Perhaps you could further say that it's exempted if
all it does is provide a well-known protocol. So a web server would
be excepted, but not RPC access to proprietary functions.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Mark Rafn

2003-08-08 05:38:47 UTC

Post by Jeremy Hankins
Email isn't entirely electronic unless it's also automatic. If you
type in the message and send it, there's a decidedly non-electronic
(well, non-digital) element: you.

Ok, so as long as someone presses a button as part of the process, this
requirement doesn't kick in? Does this mean moderated lists are
exempt and unmoderated ones would have to provide source and configuration
to all users?

Post by Mark Rafn
It's a restriction on use (per definition 1.4 section b). DFSG has
no explicit item that use of the software must not be restricted,
but any use restriction completely breaks users' trust of the
freeness of Debian.

This seems like a very difficult argument to know when to make.

I think it's one that should be made when use restrictions, which normally
fall outside of copyright law, are part of a license under consideration.

Post by Jeremy Hankins
Do you apply it to any new license, suggesting that any license not
already in Debian will break users' trust of Debian's freeness?

I apply it to old licenses as well. If Debian includes software with
restricted use, I believe it is a bug.

Post by Mark Rafn
Completely different. GPL is about distribution, and specifically says
that no use of the software is restricted. APSL limits use.

So? This is a serious question: why does that matter?

Hmm. I'll need to put more thought into how to explain this - it's the
first time I've seen it challenged. Off the cuff: freedom to use a piece
of software is more fundamental than the freedom to modify or distribute
it.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-08 13:59:18 UTC

Post by Jeremy Hankins
Email isn't entirely electronic unless it's also automatic. If you
type in the message and send it, there's a decidedly non-electronic
(well, non-digital) element: you.

I dunno. Possibly. Frankly, this case doesn't interest me much; I
can see it going either way and both are fairly reasonable. Why does
it matter?

This seems like a very difficult argument to know when to make.

I think it's one that should be made when use restrictions, which normally
fall outside of copyright law, are part of a license under consideration.

In the case of the DPSL that's not obvious, since they seem to want to
include restrictions on performance. My interpretation of that is
that they think that the "provide a service" variety of "external
deployment" would qualify as a performance -- or so they would argue,
at any rate.

Certainly at one point the argument against use restrictions was made,
but it was because in order to enforce it you'd need something like a
clickwrap rather than simply a license. For reasons that have been
discussed before, the clickwrap would inevitably make the license
non-free.

But the APSL doesn't do that, that I saw on my quick read-through. So
my understanding (IANAL, of course) is that unless the type of use
they seek to control is prohibited by copyright law (i.e., as public
performance) it has no force anyway.

So if you're right, and this kind of use it outside of copyright law,
it has no force. Otherwise, it is under copyright law. Of course, I
have no idea how this would work outside the US.

Post by Mark Rafn
Completely different. GPL is about distribution, and
specifically says that no use of the software is restricted.
APSL limits use.

So? This is a serious question: why does that matter?

Hmm. I'll need to put more thought into how to explain this - it's
the first time I've seen it challenged. Off the cuff: freedom to
use a piece of software is more fundamental than the freedom to
modify or distribute it.

The one argument against it that I know of (described above) doesn't
apply in this case. Also note that, as I said before, the kind of use
isn't restricted, only that if you use it under certain circumstances
you must also do something else (distribute source).

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Mark Rafn

2003-08-08 16:36:50 UTC

Post by Mark Rafn
Ok, so as long as someone presses a button as part of the process, this
requirement doesn't kick in? Does this mean moderated lists are
exempt and unmoderated ones would have to provide source and configuration
to all users?

I dunno. Possibly. Frankly, this case doesn't interest me much; I
can see it going either way and both are fairly reasonable. Why does
it matter?

Partly because I'm curious where you draw the line, and partly to show at
least one of the the problems with such a requirement. It only kicks in
when the system is used for what computers are good at - automating
repetitive tasks.

This seems like a very difficult argument to know when to make.

I think it's one that should be made when use restrictions, which normally
fall outside of copyright law, are part of a license under consideration.

In the case of the DPSL that's not obvious, since they seem to want to
include restrictions on performance.

This is interesting, and AFAIK the first license Debian has
considered which makes such a claim (that allowing someone to send input
to and recieve output from a program is public performance).

Post by Jeremy Hankins
Certainly at one point the argument against use restrictions was made,
but it was because in order to enforce it you'd need something like a
clickwrap rather than simply a license.

I read it in reverse. One of the main reasons for a click-through is to
limit use (as opposed to copying), and since this can never be free,
there's no reason to have a click-through on free software.

Post by Jeremy Hankins
But the APSL doesn't do that, that I saw on my quick read-through. So
my understanding (IANAL, of course) is that unless the type of use
they seek to control is prohibited by copyright law (i.e., as public
performance) it has no force anyway.

It'll be fun to see if it ever gets tested. I don't know of any other
software license that attempts to put such a limit without explicitly
claiming that use itself is covered by the license (usually by a
click-through or shrink-wrap license, but sometimes just by including a
license with the software).

I'll be semi-unavailable for a week or so, further replies (from me; I
hope others will continue to participate as new points come up) will be
sporadic.
--
Mark Rafn ***@dagon.net <http://www.dagon.net/>

Jeremy Hankins

2003-08-08 17:55:39 UTC

Post by Jeremy Hankins
In the case of the DPSL that's not obvious, since they seem to want to
include restrictions on performance.

This is interesting, and AFAIK the first license Debian has
considered which makes such a claim (that allowing someone to send
input to and recieve output from a program is public performance).

That isn't precisely the claim they're making. In fact, I don't think
they explicitly make any claim that such-and-such activity is public
performance. What they do say:

2.2 Modified Code. You may modify Covered Code and use,
reproduce, display, perform, internally distribute within Your
organization, and Externally Deploy Your Modifications and
Covered Code, for commercial or non-commercial purposes, provided
that in each instance You also meet all of these conditions:

And then condition (c) includes the requirement that source be
available. So at no point (unless I missed it -- always possible) do
they claim that external deployment (or even providing a service) is
performance. The way I interpret that they feel they can argue that
it is in court.

Of course, they also permit "use" on those condition as well, so I may
be misunderstanding their intention entirely. Maybe they think they
can get away with restricting use outside of the scope of copyright
law in the license. Maybe they even can, I certainly don't know.

Well, either way, it doesn't argue for use restrictions necessarily
being non-free in this case.

As I understand the argument to date, that seems to be the crucial
issue needed to decide whether the APSL is simply non-free and beyond
hope of being otherwise, or whether a few clarifications on the scope
of what qualifies as providing a service would suffice to make it
free. Does anyone disagree with that summary?

Hrm. On second look, there may be a few other objectionable bits to
the APSL. Section 6 says, speaking of offering support and/or
indemnity:

You must obtain the recipient's agreement that any such Additional
Terms are offered by You alone, and You hereby agree to indemnify,
defend and hold Apple and every Contributor harmless for any
liability incurred by or claims asserted against Apple or such
Contributor by reason of any such Additional Terms.

Does that mean that if your customer decides for whatever reason to
sue Apple over the code, you have to defend them?

And then section 11 Apple also asserts that they can take your
modifications and relicense them however they like. Which I think
makes it non-free (as you're required to compensate Apple with an
open-ended license in return for the right to modify), but IIRC others
here have disagreed with me on that. At the least it's really, really
slimy.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Anthony DeRobertis

2003-08-10 10:40:59 UTC

Post by Jeremy Hankins
You must obtain the recipient's agreement that any such Additional
Terms are offered by You alone, and You hereby agree to indemnify,
defend and hold Apple and every Contributor harmless for any
liability incurred by or claims asserted against Apple or such
Contributor by reason of any such Additional Terms.
Does that mean that if your customer decides for whatever reason to
sue Apple over the code, you have to defend them?

"reason of any such Additional Terms." (Without reading the rest of the
license) if I offer a warranty, I must indemnify, defend, etc. Apple and
everyone else from claims against my warranty.

Post by Jeremy Hankins
And then section 11 Apple also asserts that they can take your
modifications and relicense them however they like.

Huh? All it seems to speak of is licensing Apple's code however it
wants.

Jeremy Hankins

2003-08-12 00:50:51 UTC

Post by Anthony DeRobertis

"reason of any such Additional Terms." (Without reading the rest of the
license) if I offer a warranty, I must indemnify, defend, etc. Apple and
everyone else from claims against my warranty.

I should have seen that. Never mind.

Post by Anthony DeRobertis

Post by Jeremy Hankins
And then section 11 Apple also asserts that they can take your
modifications and relicense them however they like.

Huh? All it seems to speak of is licensing Apple's code however it
wants.

Again, my mistake.

I'll go off & hide now. :P

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Anthony DeRobertis

2003-08-10 10:37:03 UTC

Post by Jeremy Hankins
In the case of the DPSL that's not obvious, since they seem to want to
include restrictions on performance. My interpretation of that is
that they think that the "provide a service" variety of "external
deployment" would qualify as a performance -- or so they would argue,
at any rate.

"To ''perform'' a work means to recite, render, play, dance,
or act it, either directly or by means of any device or
process or, in the case of a motion picture or other
audiovisual work, to show its images in any sequence or to
make the sounds accompanying it audible." (Title 17, Sec. 101)

So, who wants to act out Quicktime Streaming Server?

Nathanael Nerode

2003-08-08 06:28:29 UTC

This seems like a very difficult argument to know when to make. Do
you apply it to any new license, suggesting that any license not
already in Debian will break users' trust of Debian's freeness?

Restrictions on use are contrary to the FSF's four freedoms #0. I think
use restrctions are a *very* reasonable thing to prohibit given that
"Debian will remain 100% Free Software".

--
Nathanael Nerode <neroden at gcc.gnu.org>
http://home.twcny.rr.com/nerode/neroden/fdl.html

Jeremy Hankins

2003-08-08 14:05:29 UTC

Post by Nathanael Nerode
Restrictions on use are contrary to the FSF's four freedoms #0. I think
use restrctions are a *very* reasonable thing to prohibit given that
"Debian will remain 100% Free Software".

Not at all:

Freedom 0: The freedom to run the program, for any purpose.

As I read it that refers to restrictions on the type of use, which is
not what's happening in the APSL. You can still run APSL 2.0 licensed
works for whatever purpose you like, but if you do so (and your use
qualifies as external deployment) you must also distribute source.

Though it's far from conclusive, it's worth noting that RMS clearly
doesn't think the APSL runs afoul of freedom 0.

--
Jeremy Hankins <***@nowan.org>
PGP fingerprint: 748F 4D16 538E 75D6 8333 9E10 D212 B5ED 37D0 0A03

Brian Kimball

2003-08-07 18:07:14 UTC

Post by MJ Ray
It is expected that GPL-3 will contain something similar to the Affero GPL
requirement for remote services to offer users the code.

We should petition the FSF to go all the way and require a guarantee of
full write access to the machines providing these services.

After all, I can't fix a google bug if I can't apply my patch to google,
right?

[Those bastards. How dare they keep me off their machines!!!]

MJ Ray

2003-08-07 18:58:39 UTC

Post by Brian Kimball
We should petition the FSF to go all the way and require a guarantee of
full write access to the machines providing these services.

I think that you have broken normal logical extension in two emails about
this licence now. Why do you think that offering a fixed service version
yourself is not a viable option?

Brian Kimball

2003-08-07 20:19:06 UTC

Post by Brian Kimball
We should petition the FSF to go all the way and require a guarantee of
full write access to the machines providing these services.

I think that you have broken normal logical extension in two emails about
this licence now.

That was my first post on this topic. You've got me confused with
someone else. Or am I not parsing your sentence correctly?

Oh, and I was joking.

Post by MJ Ray
Why do you think that offering a fixed service version
yourself is not a viable option?

Because it's all about the content, not the program delivering the
content. Running scoop on my own server won't fix any problems I might
have with k5.

MJ Ray

2003-08-07 21:53:23 UTC

Post by Brian Kimball
That was my first post on this topic. You've got me confused with
someone else. [...]

Quite likely. Apologies. Hence the vague terms.

Post by Brian Kimball
Oh, and I was joking.

It is remarkable that your jokes are not dissimilar from the posts of
someone else ;-)

[...]

Post by Brian Kimball
Because it's all about the content, not the program delivering the
content. [...]

And so we fold into another thread about licensing of other electronic
works. Should we have similar disclosure in all electronic works?

--
MJR/slef My Opinion Only and possibly not of any group I know.
http://mjr.towers.org.uk/ jabber://***@jabber.at

Anthony DeRobertis

2003-08-07 19:52:47 UTC

Post by MJ Ray
It is expected that GPL-3 will contain something similar to the Affero GPL
requirement for remote services to offer users the code. Do you object
to that? If so, why?

Depending on exactly what it is, probably. For example, if I were to
use a (hypothetical) GPLv3-covered firewall, would I have to offer my
firewall code --- rules and all, gotta have complete source --- for
download, just because it routed the packets?

I could maybe be convinced to call it OK for some very limited
circumstances involving highly-integrated network applications. I, in
general, have a problem with requiring me to distribute source when
I've never done any copying, distributing, etc.

The only thing I've done is private modifications, which could very
likely not require copyright licenses to do anyway (they could be fair
use).

Henning Makholm

2003-08-08 01:01:28 UTC