Possible weakness in GNU copyleft?
Richard Stallman
But there is a problem....... Imagine this.....
1) We distribute our web software under the GNU copyleft
2) Big company X, takes our software adds extensions to it, builds a
web site around it, and they puts our site out of business.
It is definitely possible for someone to make a modified version of a
free program, and not release it.
However, it would be misleading to describe this as a problem with the
GNU GPL in particular, because the same is true for any free software
license, including the MIT Scheme license that you use now.
In some cases, I would agree that it is antisocial to make a useful
modification and not release it. But if a license made that illegal,
if a license did not allow people to make and use their own private
changes, without publishing them, I would consider the license
unacceptably restrictive. People should always have the right to
share with their neighbors, but compelling people to share can be
wrong, especially when it conflicts with privacy rights.
Speaking practically, if someone wants to write software to do some
special job on a web site, and never release this software, he can
probably do it in a separate program which would not be an extension
of your program, and not covered by your license. For example, he
could write a Perl script (as so many webmasters do), and fork Perl to
run it. That might be kludgy, but since the software will not be
released, he probably won't consider that much of a problem.
So I'd suggest that you not try to write a restrictive license that
would take away privacy for changes. The solution would cause a worse
problem than the one it solves. It's better to use an ordinary free
software license.
William C. Cheng
Richard Stallman <r...@gnu.org> wrote:
>You wrote:
>
> But there is a problem....... Imagine this.....
>
> 1) We distribute our web software under the GNU copyleft
>
> 2) Big company X, takes our software adds extensions to it, builds a
> web site around it, and they puts our site out of business.
>
>It is definitely possible for someone to make a modified version of a
>free program, and not release it.
>
>However, it would be misleading to describe this as a problem with the
>GNU GPL in particular, because the same is true for any free software
>license, including the MIT Scheme license that you use now.
If I contribute to a GPL'ed software, I would assume that my code and
all copies of my code remains free because it's under GPL. But I would
be wrong in this case because some copies of my code can become captive.
Therefore, I would consider this to be a problem with GPL (it's probably
an inherient problem, but still a problem). No other free software
license makes the claim that my code will remain free if I use it.
>In some cases, I would agree that it is antisocial to make a useful
>modification and not release it. But if a license made that illegal,
>if a license did not allow people to make and use their own private
>changes, without publishing them, I would consider the license
>unacceptably restrictive.
Yes. It's perfectly reasonable to allow modifications for personal use.
Most free software developers probably won't mind if their code become
captive on these machines.
>People should always have the right to
>share with their neighbors, but compelling people to share can be
>wrong, especially when it conflicts with privacy rights.
I thought if one uses GPL, one is compelled to share! (Of course, I
don't have to use GPL.) Can someone kindly explain how "compelling
people to share can be wrong" is consistent with GPL!?
>Speaking practically, if someone wants to write software to do some
>special job on a web site, and never release this software, he can
>probably do it in a separate program which would not be an extension
>of your program, and not covered by your license. For example, he
>could write a Perl script (as so many webmasters do), and fork Perl to
>run it. That might be kludgy, but since the software will not be
>released, he probably won't consider that much of a problem.
>
>So I'd suggest that you not try to write a restrictive license that
>would take away privacy for changes. The solution would cause a worse
>problem than the one it solves. It's better to use an ordinary free
>software license.
Why can't we just say that GPL has this problem and that putting your
software under GPL doesn't guarantee its complete freedom!
--
Bill Cheng // bill....@acm.org <URL:http://bourbon.cs.umd.edu:8001/william/>
Barry Margolin
William C. Cheng <wil...@cs.umd.edu> wrote:
>I thought if one uses GPL, one is compelled to share! (Of course, I
>don't have to use GPL.) Can someone kindly explain how "compelling
>people to share can be wrong" is consistent with GPL!?
You thought wrong. The GPL puts conditions on how you can redistribute the
code: if you redistribute the binary, you must also make the source
available. It doesn't force you to redistribute in the first place,
though. The purpose of the GPL is to ensure that someone can modify the
program they receive.
--
Barry Margolin, bar...@bbnplanet.com
GTE Internetworking, Powered by BBN, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Don't bother cc'ing followups to me.
Eleftherios Gkioulekas
> In article <1999011405...@psilocin.gnu.org>,
> Richard Stallman <r...@gnu.org> wrote:
> >You wrote:
> >
> > But there is a problem....... Imagine this.....
> >
> > 1) We distribute our web software under the GNU copyleft
> >
> > 2) Big company X, takes our software adds extensions to it, builds a
> > web site around it, and they puts our site out of business.
> >
> >It is definitely possible for someone to make a modified version of a
> >free program, and not release it.
> >
> >However, it would be misleading to describe this as a problem with the
> >GNU GPL in particular, because the same is true for any free software
> >license, including the MIT Scheme license that you use now.
>
> If I contribute to a GPL'ed software, I would assume that my code and
> all copies of my code remains free because it's under GPL. But I would
> be wrong in this case because some copies of my code can become captive.
> Therefore, I would consider this to be a problem with GPL (it's probably
> an inherient problem, but still a problem). No other free software
> license makes the claim that my code will remain free if I use it.
The GPL has the specific goal of creating a reality equivalent to one
in which there is no copyright law.
Without copyright law, everyone would have the right to use, modify
and distribute software. It would not be possible to impose proprietary
restrictions and criminalize those who make copies, because there would
be no law that says that you can't make copies. However, it would also
not be possible to require people to publish their modifications.
The GPL is essentially an "emulator" to this reality.
> >In some cases, I would agree that it is antisocial to make a useful
> >modification and not release it. But if a license made that illegal,
> >if a license did not allow people to make and use their own private
> >changes, without publishing them, I would consider the license
> >unacceptably restrictive.
>
> Yes. It's perfectly reasonable to allow modifications for personal use.
> Most free software developers probably won't mind if their code become
> captive on these machines.
>
> >People should always have the right to
> >share with their neighbors, but compelling people to share can be
> >wrong, especially when it conflicts with privacy rights.
>
> I thought if one uses GPL, one is compelled to share! (Of course, I
> don't have to use GPL.) Can someone kindly explain how "compelling
> people to share can be wrong" is consistent with GPL!?
You are compelled to share, when you decide to share at all. So when you
decide to distribute your program, you have to give the source code
as well. You can however decide not to share and keep your copy private.
Freedom includes both the freedom not to share and the freedom to share.
The freedom not to share stuff with everyone else is called "privacy".
> >Speaking practically, if someone wants to write software to do some
> >special job on a web site, and never release this software, he can
> >probably do it in a separate program which would not be an extension
> >of your program, and not covered by your license. For example, he
> >could write a Perl script (as so many webmasters do), and fork Perl to
> >run it. That might be kludgy, but since the software will not be
> >released, he probably won't consider that much of a problem.
> >
> >So I'd suggest that you not try to write a restrictive license that
> >would take away privacy for changes. The solution would cause a worse
> >problem than the one it solves. It's better to use an ordinary free
> >software license.
>
> Why can't we just say that GPL has this problem and that putting your
> software under GPL doesn't guarantee its complete freedom!
The GPL does guarantee complete freedom, including freedom to privacy.
What it does not guarantee is complete availability. You can't have complete
freedom and complete availability at the same time.
Also, as RMS is pointing out, even if you craft a license that forces
availability, it can still be defeated. So the point is moot.
Elef.
--
Eleftherios Gkioulekas | "The truth points to itself"
http://www.amath.washington.edu/~lf/ | --Kosh
Joe Buck
>all copies of my code remains free because it's under GPL. But I would
>be wrong in this case because some copies of my code can become captive.
I can't think of any "solution" to this that wouldn't cause more problems.
For example, let's suppose you said that if anyone modifies your code
they must distribute the modified version. OK, I open up my editor and
start modifying. I've typed one character. I now have a modified version
that must be published. OK, so you say I don't have to publish it until
I use the program internally. OK, my mods compile for the first time and
I test it, but the program is in an incoherent state. I wouldn't want
to publish it and no one would want it until I clean it up.
Now, perhaps you can come up with some solution for this, but demanding
that you can get all changes the instant they are made will just mean
that no one can change your software.
--
-- Joe Buck
work: jb...@synopsys.com, otherwise jb...@welsh-buck.org or jb...@best.com
http://www.welsh-buck.org/
Joe Buck
>The GPL has the specific goal of creating a reality equivalent to one
>in which there is no copyright law.
I don't think that this is really the case. In the absence of copyright
law, I could distribute binaries and "hoard" source. I couldn't block
anyone from redistributing, patching, or reverse-engineering the binary,
but I could attempt to exploit my special access to the source to
increase the amount of consulting money I could make, since I'd be the
only one in a position to quickly make improvements to the program.
brl...@my-dejanews.com
Eleftherios Gkioulekas <l...@amath.washington.edu> wrote:
> Without copyright law, everyone would have the right to use, modify
> and distribute software. It would not be possible to impose proprietary
> restrictions and criminalize those who make copies, because there would
> be no law that says that you can't make copies. However, it would also
> not be possible to require people to publish their modifications.
This is not true. Without copyright, there would still be contracts. I can
say, "I'll give you A if you give me B and agree to do X and not do Y." If
you accept that offer, it's a contract and you're bound to it. So if I say,
"I'll give you this software if you don't share it with anyone else" and you
accept the offer, I can sue you if you share it with someone else, completely
apart from copyright.
What copyright adds is an implicit contract. You can buy a CD from your
local music store without signing anything or verbally agreeing not to make
copies. Nonetheless you are bound not to copy the CD except under fair use.
In a world without copyright law, you could still have non-free CDs, but
you'd have to sign a form before you could get them.
Copyrighted works may come with a "license" that allows you to make copies
under certain conditions in addition to fair use. Software sometimes comes
with something that says "license" on it but really is just a piece of trash
that claims to take away some of your fair use rights, e.g. letting a friend
use your copy. Some software publishers seem to think they can use copyright
law to bind you to an implicit contract under whatever terms they choose, but
really the only thing they can restrict is non fair-use copying unless you
agree to other restrictions. Breaking open a shrink-wrap does not constitute
agreement, even if a so-called "license" says it does.
The GPL comes into play only when you copy software. It says it applies when
you "modify or copy", but I don't think it can make stipulations on someone
who only modifies it without making non fair-use copies, any more than a
publishing company could sue me for using a pen and white-out on a book I
bought.
If you were to use a modified GPL that said, "you must share modifications",
I don't think you could enforce it unless people distributed modified copies.
For example, party A could publish a CD with your software unmodified. The
party B buys a CD and copies it to their hard drive (fair use). They can
modify it on their hard drive to their hearts' content and not be bound by
the license.
I agree with RMS that it's best to use the GPL as it stands.
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
William C. Cheng
>William C. Cheng <wil...@cs.umd.edu> wrote:
>>If I contribute to a GPL'ed software, I would assume that my code and
>>all copies of my code remains free because it's under GPL. But I would
>>be wrong in this case because some copies of my code can become captive.
>
>I can't think of any "solution" to this that wouldn't cause more problems.
I think I simply would like to see an admission of the problem instead of
denials!
>For example, let's suppose you said that if anyone modifies your code
>they must distribute the modified version. OK, I open up my editor and
>start modifying. I've typed one character. I now have a modified version
>that must be published. OK, so you say I don't have to publish it until
>I use the program internally. OK, my mods compile for the first time and
>I test it, but the program is in an incoherent state. I wouldn't want
>to publish it and no one would want it until I clean it up.
You won't ``distribute'' your code at this point either. The problem
occurs when you starting ``distributing'' source or binaries. If you
get to a point where you can run it behind a web server, you are running
the code on behalf of your users. At this point, if your user demand
source code, you should give it up! But this is not the case with GPL
because invoking a GPL'ed program from a non-GPL'ed program (using
something like system()) is allowed.
>Now, perhaps you can come up with some solution for this, but demanding
>that you can get all changes the instant they are made will just mean
>that no one can change your software.
GPL mentioned about giving users freedom. This should include running
it directly or indirectly. I would think running an internal version
within a company is consistent with the spirit of GPL. (Given that if
a user of the internal network asks for source, you will give it to
him/her.)
The solution is to consider running such code indirectory a binary
distribution. If that's too much trouble and there's no fix, I'd like
to see an acknowledgement of the problem (in the form of ``developer
beware'')!
William C. Cheng
Eleftherios Gkioulekas <l...@amath.washington.edu> wrote:
>> In article <1999011405...@psilocin.gnu.org>,
>> Richard Stallman <r...@gnu.org> wrote:
>> >People should always have the right to
>> >share with their neighbors, but compelling people to share can be
>> >wrong, especially when it conflicts with privacy rights.
>>
>> I thought if one uses GPL, one is compelled to share! (Of course, I
>> don't have to use GPL.) Can someone kindly explain how "compelling
>> people to share can be wrong" is consistent with GPL!?
>
>You are compelled to share, when you decide to share at all. So when you
>decide to distribute your program, you have to give the source code
>as well. You can however decide not to share and keep your copy private.
>Freedom includes both the freedom not to share and the freedom to share.
>The freedom not to share stuff with everyone else is called "privacy".
Isn't running binaries behind a web server sharing?
I would have no problem is someone simply runs it internally in a private
network (as long as the source code is availabe to any user of the private
network). When I contribute to a GPL'ed source, I understand that this
can happen.
>> Why can't we just say that GPL has this problem and that putting your
>> software under GPL doesn't guarantee its complete freedom!
>
>The GPL does guarantee complete freedom, including freedom to privacy.
>What it does not guarantee is complete availability. You can't have complete
>freedom and complete availability at the same time.
If my code is captive, I won't call that ``complete freedom''.
If I contribute to a GPL'ed source and a company modified it and run it
behind a web server and later I become a user my code because I access
their web server in a certain way, I want the company to free my code!!!
Is that unreasonable? Can you call this ``complete freedom''?
>Also, as RMS is pointing out, even if you craft a license that forces
>availability, it can still be defeated. So the point is moot.
I don't want a license that forces availability. As a developer, I want
to hear that, "there are loopholes in GPL where people can take advantage
of... some of your code may become captive... etc." Then I can choose to
develop under GPL or not. I want honesty. I don't want empty promises
of complete freedom!
Joe Buck
> >I can't think of any "solution" to this that wouldn't cause more problems.
>
>I think I simply would like to see an admission of the problem instead of
>denials!
Why? Such an "admission" is not in the interest of the FSF. Why should
they publish statements telling people how to exploit their code without
honoring their principles?
>You won't ``distribute'' your code at this point either. The problem
>occurs when you starting ``distributing'' source or binaries.
The GPL says you must provide source when you provide a binary.
> If you
>get to a point where you can run it behind a web server, you are running
>the code on behalf of your users. At this point, if your user demand
>source code, you should give it up!
In *your* opinion. You could write a license for your own code that
added this requirement (but you couldn't use any GPL code in your program).
>The solution is to consider running such code indirectory a binary
>distribution. If that's too much trouble and there's no fix, I'd like
>to see an acknowledgement of the problem (in the form of ``developer
>beware'')!
Again, why? Anyone who reads the GPL can determine this feature.
Better just to tell folks to make an effort to understand the license
before applying it to new software. In the past, gnu.misc.discuss has
gotten complaints from people who used the GPL without knowing what it
meant.
Barry Margolin
>Eleftherios Gkioulekas <l...@amath.washington.edu> wrote:
>>The GPL has the specific goal of creating a reality equivalent to one
>>in which there is no copyright law.
>
>I don't think that this is really the case. In the absence of copyright
>law, I could distribute binaries and "hoard" source.
Eleftherios's statement was a very simplistic summary of the GPL's goal.
That goal refers to the perspective of someone who *wants* to share --
copyright laws prohibit him from doing so (he has to get explicit
permission from the copyright holder). Without copyrights, anyone who
wants to share is free to do so. Remember, there are clauses of the GPL
that apply to the binaries as well: if you distribute a binary to someone,
you can't prohibit them from giving it away to someone else (the GPL is
more than just "open *source*").
The feature of the GPL that forces sources to be included is above and
beyond the "no copyright" ideal.
Jeffrey B. Siegal
> >all copies of my code remains free because it's under GPL. But I would
> >be wrong in this case because some copies of my code can become captive.
>
> they must distribute the modified version.
Here's a better solution.
The current GPL states that (paraphrasing) "if you distribute object
code, you must make source code available." Change that to "if you
distribute object code, or use the software to provide a computer-based
service, you must make source code available."
The question of how to define a "provide a computer-based service" is
not necessarily trivial but is a solved problem. Many (real, not
shrink-wrap) software licenses prohibit it, or require additional
license fee payments.
Andy Isaacson
>Here's a better solution.
>
>The current GPL states that (paraphrasing) "if you distribute object
>code, you must make source code available." Change that to "if you
>distribute object code, or use the software to provide a computer-based
>service, you must make source code available."
>
>The question of how to define a "provide a computer-based service" is
>not necessarily trivial but is a solved problem. Many (real, not
>shrink-wrap) software licenses prohibit it, or require additional
>license fee payments.
But such a license prohibits me from using flex/bison on an unfree
project. It seems wrong to me to prohibit such use. After all, I
didn't necessarily write the grammar in question for use with
flex/bison; I may have used AT&T yacc originally.
-andy
Barry Margolin
Jeffrey B. Siegal <j...@quiotix.com> wrote:
>The question of how to define a "provide a computer-based service" is
>not necessarily trivial but is a solved problem. Many (real, not
>shrink-wrap) software licenses prohibit it, or require additional
>license fee payments.
If I were a professional tax preparer, and I used a computer to generate
the tax returns for my clients, would I be "providing a computer-based
service"? I'm certainly providing a service, and I'm using a computer to
do it. If I used any software licensed under this variant of GPL as part
of this process, would I have to make it available to all my clients? If I
didn't modify the software, I suppose I could just give them a pointer to
where they could get it themselves (I think the clause that currently
allows this for certain binary distributions should be amended to allow
this in your suggested variant), but if I made any customizations I would
have to go into the software distribution business, even though I'm not
even in the software business to begin with.
Jeffrey B. Siegal
>
> In article <36A1322E...@quiotix.com>,
> Jeffrey B. Siegal <j...@quiotix.com> wrote:
> >The question of how to define a "provide a computer-based service" is
> >not necessarily trivial but is a solved problem. Many (real, not
> >shrink-wrap) software licenses prohibit it, or require additional
> >license fee payments.
>
> If I were a professional tax preparer, and I used a computer to generate
> the tax returns for my clients, would I be "providing a computer-based
> service"?
Not in the way I intended that term to be used. I'm not suggesting that
I know exactly how to write this definition, only that I am quite
confident that it can be done. What I had in mind is closer to
providing access to the software without actually providing a copy of
the software.
Your tax preparer example takes this idea off in a direction where the
loophole is probably not very serious (though I'm not completely sure of
that either). However, let me take in in a direction where the problem
is very serious indeed (and closer to the original discussion which
started this thread).
Imagine someone makes some important and useful improvements to, say,
gcc, but rather than distribute the binary, which would require him to
both distribute the source and allow further redistribution of both the
binary and source, he makes the improved version available over the
Internet (yeah, I know gcc and rsh won't work exactly this way, but it
is for illustration):
rsh rsh.improved-gcc.com igcc - <foo.c >foo.o
Whoever did this could even charge users for a password. There would be
no requirement to ever share either the binary or the source code to the
gcc improvements. If this charge is for a lifetime of access, this
looks virtually identical to buying a binary software package and not
being able to receive source or further redistribute the software.
Remember, most software is licensed, not sold (at least this is the
claim--ignore for the moment the weaknesses of shrink-wrap licenses).
What is the significant difference between licensing the right to use a
copy of software and licensing the right to access the software running
on a computer maintained by the licensor. What if the licensor installs
the computer at the licensee's premises, but never transfers ownership
of the computer, just licensing the customer to use it? What if the
entire computer is a single chip, and the single chip is installed into
the customer's computer, again never transferring ownership of the chip,
just licensing its use?
Ultimately this does not differ significantly from turning a modified
version of gcc into a proprietary product.
Jeffrey B. Siegal
> But such a license prohibits me from using flex/bison on an unfree
> project.
How so? It says nothing about the output of the software. It simply
talks about incorporating the software into a service. See my later
post which clarifies this as referring to allowing _access_ to the
software.
Eleftherios Gkioulekas
> In article <77m91d$68$1...@shell17.ba.best.com>, Joe Buck <jb...@best.com> wrote:
> >Eleftherios Gkioulekas <l...@amath.washington.edu> wrote:
> >>The GPL has the specific goal of creating a reality equivalent to one
> >>in which there is no copyright law.
> >
> >I don't think that this is really the case. In the absence of copyright
> >law, I could distribute binaries and "hoard" source.
>
> Eleftherios's statement was a very simplistic summary of the GPL's goal.
> That goal refers to the perspective of someone who *wants* to share --
> copyright laws prohibit him from doing so (he has to get explicit
> permission from the copyright holder). Without copyrights, anyone who
> wants to share is free to do so. Remember, there are clauses of the GPL
> that apply to the binaries as well: if you distribute a binary to someone,
> you can't prohibit them from giving it away to someone else (the GPL is
> more than just "open *source*").
>
> The feature of the GPL that forces sources to be included is above and
> beyond the "no copyright" ideal.
Which is a very important detail that, believe it or not, didn't cross
my mind all this time! Thank you both for pointing it out to me.
I have been misled by the statement "software should not have owners".
Actually, that may not necessarily imply that software should not be
covered by copyright law at all. Evidently some kind of law is needed
to make copyleft work, even in the complete absense of copyright laws.
At some other writing RMS has explained that one thing he wants from
copyright law is to recognize the concept that there is code that
belongs to the public, and not to a single owner, so that a scheme
like copyleft can be enforced by any member of the public, and not
only the owner.
Elef.
Alan Curry
Support for Compiler On A Chip (EXPERIMENTAL) ? (CONFIG_PCI_GCC) [Y/n/m] _
Maybe this is where I'd draw the line. Surely installing a computer at a
client's site is an act that qualifies as *distributing* all the software
contained in it. Of course I should mention that I haven't been tainted by
reading some lawyer's definition of the word "distribute", so I still have
some common sense.
--
Alan Curry |Declaration of | _../\. ./\.._ ____. ____.
pac...@cqc.com|bigotries (should| [ | | ] / _> / _>
--------------+save some time): | \__/ \__/ \___: \___:
Linux,vim,trn,GPL,zsh,qmail,^H | "Screw you guys, I'm going home" -- Cartman
Jeffrey B. Siegal
> Support for Compiler On A Chip (EXPERIMENTAL) ? (CONFIG_PCI_GCC) [Y/n/m] _
>
> Maybe this is where I'd draw the line. Surely installing a computer at a
> client's site is an act that qualifies as *distributing* all the software
> contained in it.
I don't think I agree with that, but even if it were true, then we're left with:
Support for Compiler On Internet (EXPERIMENTAL, $$$) ? (CONFIG_COIN_GCC) [Y/n/m]
_
Imagine a not-so-distant future where the Internet becomes truly pervasive and
where every computer (except specialized ones for classified work) includes
constant wireless Internet connectivity. (In such a world, the idea of using a
compuer without being connected would seem about as strange as the idea of using
a batch-oriented computer driven by punched cards seems today.)
In such a world, what's the difference between an application installed locally,
which has been "distributed" to you, and and one you use remotely without ever
receiving a copy?
William C. Cheng
>William C. Cheng <wil...@cs.umd.edu> wrote:
>> >I can't think of any "solution" to this that wouldn't cause more problems.
>>
>>denials!
>
>Why? Such an "admission" is not in the interest of the FSF. Why should
>they publish statements telling people how to exploit their code without
>honoring their principles?
FSF likes to recuite programmers (I'm not talking about employment) to
contribute to GPL'ed code. Like the previous poster, programmers may get
the impression that their contributed code will have ``complete freedom''.
I understand that FSF needs marketing, but misinforming programmers about
what they are getting from GPL is, well, not nice.
>>You won't ``distribute'' your code at this point either. The problem
>>occurs when you starting ``distributing'' source or binaries.
>
>The GPL says you must provide source when you provide a binary.
>
>> If you
>>get to a point where you can run it behind a web server, you are running
>>the code on behalf of your users. At this point, if your user demand
>>source code, you should give it up!
>
>In *your* opinion. You could write a license for your own code that
>added this requirement (but you couldn't use any GPL code in your program).
FSF has always try to convince/urge people to use GPL.
>>The solution is to consider running such code indirectory a binary
>>distribution. If that's too much trouble and there's no fix, I'd like
>>to see an acknowledgement of the problem (in the form of ``developer
>>beware'')!
>
>Again, why? Anyone who reads the GPL can determine this feature.
I've heard this argument before, and it's absolutly not true! I would
think that only a small percentage of the people reading the GPL can
figure out the subtleties. It took me many many reads and reading many
discussions in this newsgroup to determine this. The original poster
(Joseph C Wang <j...@mit.edu>) must have also read GPL many time and
wondered, "this can't be, I must be reading it wrong".
I feel that people who volunteers their time and effort and contribute
to GPL'ed code deserve a little more honesty.
>Better just to tell folks to make an effort to understand the license
>before applying it to new software. In the past, gnu.misc.discuss has
>gotten complaints from people who used the GPL without knowing what it
>meant.
That's why I'm making a little noise here. Just don't want this ``bug''
to be covered up, again!
Christopher Browne
...
In the case of tax return preparation (a good example, by the way), it
is quite clear that there is not much need to 'export' the software as
part of the service. Indeed, at a competent service, the service is
likely not really the computer, but rather the staff that hopefully ask
you the right questions about your tax situation.
>However, let me take in in a direction where the problem
>is very serious indeed (and closer to the original discussion which
>started this thread).
>
>Imagine someone makes some important and useful improvements to, say,
>gcc, but rather than distribute the binary, which would require him to
>both distribute the source and allow further redistribution of both the
>binary and source, he makes the improved version available over the
>Internet (yeah, I know gcc and rsh won't work exactly this way, but it
>is for illustration):
>
>rsh rsh.improved-gcc.com igcc - <foo.c >foo.o
>
>Whoever did this could even charge users for a password. There would be
>no requirement to ever share either the binary or the source code to the
>gcc improvements. If this charge is for a lifetime of access, this
>looks virtually identical to buying a binary software package and not
>being able to receive source or further redistribute the software.
It requires that you *trust* the producer of the "service."
In the case of the tax service, there is a product, namely a tax return,
whose quality may be fairly readily evaluated. And once filed, it
doesn't really get obsolete.
The above approach requires that you trust the producer of this
'compilation service' to continue to support the system. It would be
downright stupid to trust this for any important code if there is
nothing that dictates that the service will continue to be available
next week let alone next year.
By the time our erstwhile service provider builds up infrastructure to
establish that he'll be around in year 2000, perhaps survive to March 1
2000, and possibly even on into the next millennium, this establishes a
considerable set of costs associated with *establishing trust.*
Which makes the service rather expensive. If the service *doesn't*
prove to be expensive, I would question whether they're charging
enough...
Those costs aren't nearly so necessary with free software, because
having the sources mitigates considerably the need to trust the vendor.
Note that this same question of trust is why Brad Cox's idea of "object
licensing" never took off. He had the neat idea of having some sort of
"object broker" which would let you run someone's objects, and charge a
fee for each run. (One could integrate this into an ORB; it doesn't
appear that anyone has implemented this yet.)
Unfortunately, this means having to have a scheme to *deny* running
objects. Which introduces some wonderful new kinds of system failures:
- If you haven't got an account at the "payment broker," the code won't
run. This is an intended failure.
- If you haven't got credit at the "broker," you won't be allowed to
run the code. Also intentional.
- If you can't connect to the "payment broker," the code won't run.
Oops. Someone backhoed a T1 between you and the broker. Program
stops.
- Broker goes out of business. Oops. Program stops.
In order to have this sort of 'object broker,' you need to have an
additional set of bureaucracy, reliable networks, and need to trust a
bunch of parties in order for the system to work.
In practice, it makes sense to pay an extra few thousand dollars to get
a "broker free" license to the code, and eliminate this enormous
reliability hole. Or use free software, and make the whole issue moot.
For the important issues, it's really all about trust...
--
"Oh, I've seen copies [of Linux Journal] around the terminal room at
The Labs." (By Dennis Ritchie)
cbbr...@ntlug.org- <http://www.ntlug.org/~cbbrowne/lsf.html>
Lieven Marchand
> In article <77lil6$j...@bourbon.cs.umd.edu>,
> William C. Cheng <wil...@cs.umd.edu> wrote:
> >I thought if one uses GPL, one is compelled to share! (Of course, I
> >don't have to use GPL.) Can someone kindly explain how "compelling
> >people to share can be wrong" is consistent with GPL!?
>
> You thought wrong. The GPL puts conditions on how you can redistribute the
> code: if you redistribute the binary, you must also make the source
> available. It doesn't force you to redistribute in the first place,
> though. The purpose of the GPL is to ensure that someone can modify the
> program they receive.
One problem area here is the definition of distribution. If my
employer gives me access to a internally modified version of a GPLed
program, do I have the right to ask for the sources and redistribute
them? Can my employer forbid me to do this in my employment agreement?
And does it make a difference whether I can only use this program on
company equipment or also get a version to run on my private machine
at home?
--
Lieven Marchand <m...@bewoner.dma.be>
------------------------------------------------------------------------------
Few people have a talent for constructive laziness. -- Lazarus Long
Barry Margolin
Lieven Marchand <m...@bewoner.dma.be> wrote:
>One problem area here is the definition of distribution. If my
>employer gives me access to a internally modified version of a GPLed
>program, do I have the right to ask for the sources and redistribute
>them? Can my employer forbid me to do this in my employment agreement?
>And does it make a difference whether I can only use this program on
>company equipment or also get a version to run on my private machine
>at home?
I think most people consider distribution to be between organizations. If
it were commercial software, the vendor wouldn't expect your employer to
charge you for the software they put on your desktop, so it's clear that
this is not typically considered to be distribution.
But this doesn't seem to be at all related to the topic of this thread,
which is about situations where the software isn't distributed, but used in
the implementation of an online service (specifically, a web-based
service).
Christopher Browne
>Barry Margolin <bar...@bbnplanet.com> writes:
>> In article <77lil6$j...@bourbon.cs.umd.edu>,
>> William C. Cheng <wil...@cs.umd.edu> wrote:
>> >I thought if one uses GPL, one is compelled to share! (Of course, I
>> >don't have to use GPL.) Can someone kindly explain how "compelling
>> >people to share can be wrong" is consistent with GPL!?
>>
>> You thought wrong. The GPL puts conditions on how you can redistribute the
>> code: if you redistribute the binary, you must also make the source
>> available. It doesn't force you to redistribute in the first place,
>> though. The purpose of the GPL is to ensure that someone can modify the
>> program they receive.
>
>employer gives me access to a internally modified version of a GPLed
>program, do I have the right to ask for the sources and redistribute
>them? Can my employer forbid me to do this in my employment agreement?
>And does it make a difference whether I can only use this program on
>company equipment or also get a version to run on my private machine
>at home?
It makes sense to treat "redistribution" on an organization basis, where
redistribution means "it leaves the organization," as opposed to on a
purely technical basis, where redistribution might mean "we copied the
binaries from one disk array to another."
If the company that employs you gives you access to that modified
version, it is reasonable to interpret this as representing having
access in your role as part of the company.
In theory, you might legitimately have the right to demand access to the
sources under the GPL; suing your own employer for this would be an act
of exceedingly dubious wisdom if you plan on sticking around.
--
"DTDs are not common knowledge because programming students are not
taught markup. A markup language is not a programming language."
-- Peter Flynn <silm...@m-net.arbornet.org>
cbbr...@ntlug.org- <http://www.hex.net/~cbbrowne/lsf.html>