I understand that producing code by contract can be a way to make money. But unfortunately I can not cite such companies as successful, because they never release their damn financials! I have no proof for any company but Red Hat and MySQL.
Well, I have discussed this with several lawyers. In fact, I wrote it for the lawyers, and they paid me for it. Much of my work is being the bridge between engineering and legal departments, or the producer of Open Source policies for companies - which of course lawyers have to approve.
It sounds to me as if the license expects conveyance of source code along with the car, and you can't expect the average guy selling his car, or the average car dealer, to even know what source code is. And none of the lawyers have told me I'm full of it yet.
MySQL has worked out a successful means of being paid while producing Open Source. To do so they separate the people who want to share from those who do not, and get a fair quid-pro-quo from both. If proprietary software stopped existing this would not work, but for the forseeable future it does. Their proprietary licensed program is identical to the Open Source one (wasn't always, I know). As far as I can tell, it's all good.
Sale of an embedded device is distribution or conveyance of the software within the embedded device - even though it is not production of an additional copy. Here is the applicable GPL2 text regarding what you then must do if you distribute:
a) Accompany it with the complete corresponding machine-readable
source code, which must be distributed under the terms of Sections
1 and 2 above on a medium customarily used for software interchange; or,
b) Accompany it with a written offer, valid for at least three
years, to give any third party, for a charge no more than your
cost of physically performing source distribution, a complete
machine-readable copy of the corresponding source code, to be
distributed under the terms of Sections 1 and 2 above on a medium
customarily used for software interchange;
There is a third option in the GPL text that doesn't apply to commercial sale.
It really doesn't say you are responsible to provide the source of a modification. It says you have to provide the whole thing. It doesn't offer modification as a condition for this particular rule, just distribution.
The second and subsequent paragraphs of your argument are philosophical, and I think it's sufficient just to say I don't agree.
It's not use. It's creation of a derivative work. None of your scenario plays out until a derivative work is created.
I remember about 40 lawyers being in the room during some of the GPL3 meetings. GPL3 is the most legally-reviewed license we've seen, and the AGPL3 adds reasonably little text, which itself has had a pretty good review. I doubt it'll fall over so easily as you think.
I think we can indeed woo open hardware vendors, but I am much less confident that they can be the existing vendors. They just have too many conflicts of interest for it to work.
The first real piece of Open Hardware I've purchased is the ip08 PBX from David Rowe's store. No proprietary code at all in the box. It came with the screws that hold the box together and the rubber feet in a little zip-lock bag. In other words, the default mode of the unit when shipped is one in which you can "open the hood". No tricky process to crack the box open as with so many consumer products. And this even though the daughter cards came already installed. It's manufactured for David by Atcom, but the design is David's and is itself Open Sourced. Anyone can manufacture it.
It's using the Asterisk GUI 2.0, which is a little too fresh for the less technical, but it'll mature.
Things are looking up a bit, but you noticed the big anti-net-neutrality putsch going on today, didn't you? And although audio seems to have opened up significantly, video is progressing in the opposite direction. The next iteration of HDML allows no analog video at all.
I don't see any judge interpreting it as precedent that once you own a copy, you own all rights. RMS would be joyful if copyright were entirely gutted and the proprietary software business was put out of business, which is what that would do.
A more limited interpretation is that ownership of a copy embodies a limited right to use, period. Which doesn't bother us.
would ever in their right mind touch anything released under it.
They said that about GPL 2 as well. Now it's getting difficult to buy some kinds of devices without GPL code in them.
My experience with the Sony DHG-HDD250 has been educational. I have two of these DVRs. They have the Busybox license (I created busybox) in the user manual. The user is not permitted to set the time. It can only be set by TV-Guide-On-Screen's signal. I guess that setting the time would allow circumvention of some DRM information somewhere. I suspect they planned to send DRM info over the TV-Guide-On-Screen signal. Or they do.
So, when the digital TV cut-over happened, SONY had not updated the unit to get the digital TV-Guide-On-Screen signal. So, the unit was effectively a brick. You couldn't set the time, and there was no program guide.
Sony finally did an update, some weeks after the cut-over. But they were under no obligation to do so, as by then all the units were out of warranty. I'd spent about $1000 to buy the pair and for a while they were useless.
Next time I'm buying open hardware. And I am licensing new code under GPL3 and AGPL3.
How did this get to be the case? What has changed, that it won't remain the case?
It originally came about because of the reverse-engineering of IBM BIOS. What has changed now is that devices are built to exercise strong DRM on digital media, and DMCA includes tremendous fines for circumvention of copyright-management systems, including the copyright of the device itself and its installed software.
The easiest way for us to be locked out would be if significant sites started to require TPM and a properly-signed OS for web access. They do want to protect their content from the evil "view source" button, etc.
I suspect that Wikipedia works that way, but I don't have proof. There are a lot of retired people with deep expertise that they would like to gift to the world, and the world is generally resistant to long lectures from old folks. So, the retirees put it in Wikipedia and feel fulfilled.
I don't see a reason for software development to transition to that demographic, though. You'd need some evidence that current retirees still want to program in significant and growing numbers - that they haven't already fulfilled that urge to their satisfaction. I think you are assuming that their enthusiasms won't change as they mature. And the numbers would have to be sufficient to overwhelm the presence of younger people involved in industry.
Yes, that is what dual-licensing is about. It allows you to make Open Source, and sell it as proprietary software to folks who don't want to free up their derivative works.
In both cases there is a quid-pro-quo. It is either software, or money.
We know of only two profitable Open Source companies, MySQL and Red Hat. The others have not released financials.
It can't mean asserting control, because the OSD does not require that any rights be withheld. Instead, it requires that rights beyond the default of "All Rights Reserved" must be conveyed. Public domain conveys all possible rights.
Well, BSD and Apache are also only Open Source if there is source code. Yet, the BSD and Apache licenses appear on the OSI web site as accepted licenses. The only reason "Public Domain" is not listed there is that it is not a license.
MySQL. They were profitable, and probably still are, and Sun bought them for 1.1 Billion dollars after they'd actually prepared, but not filed, an IPO. Too bad what happened to Sun after that, but it wasn't MySQL's fault.
FSF has that here. My own rationale is that application of DMCA to my code in a locked down device makes my code essentially proprietary. I might be persuaded to write proprietary code for someone, but I'd want to be paid for that. My motivation for writing code that I'm not paid for is that the code will remain free.
Do you really think that adding a bizarre twist to what's already a complicated and confusing legal system is patching a bug?
It's a lot like firewall code. You have to deal with a hostile environment.
From your suggestion, anyone running a Linux webserver would have to make the source available per the GPL? Computer labs and internet cafes using any GPL software as well? What if I let a stranger on the bus use my laptop to check e-mail?
For one of my customers I have defined processes that will allow the casual sale of a motor vehicle with Linux in the dashboard to happen without infringing the GPL. Technically, the seller should really give the buyer source code. But they aren't expected to be computer-literate. So, the manufacturer has to provide some services on their behalf.
Nobody is proposing to apply AGPL3 to Linux. But I am applying it to some new embedded software. And yes, if you perform it to clients over a network, the source code has to be made available.
It's Open Source if you have source code and the rights, stated in more detail in the OSD, to use, redistribute, and modify. Public domain would give you all those rights. The only remaining question once a program is public domain is "Is the source code available?" If it is, the program is really Open Source.
The TCP/IP was something called "winsock" from a third party, but it was definitely an add-on to 3.1 . As far as WfW, I think that out of the box there was the early SMB-based networking, not TCP/IP. And there was Novell.
Access to source and access to hardware are two totally unrelated things.
You say that now, because you have motherboards available to you that don't even come with an operating system and are publicly documented so that you can run the OS you like. But your access to such things is by no means guaranteed. Consider a future in which embedded devices like phones are the way we use computers. Sometime in the future there could be no mass-market hardware that would allow Free Software to run. You won't be able to get the right code-signing, or a place in the app store. RMS was concerned about that. IMO for good reason. Thus the anti-circumvention material in GPL3.
By the way, we only get away with manufacturing hardware that runs Free Software for software-defined radio today because it's "test equipment". FCC has no intention of allowing open hardware for radio applications to reach the mass market. We could get into a similar situation with other open hardware.
If RMS feels that way, why didn't he merge the AGPL into the GPLv3?
Why doesn't he make LGPL into GPL? He gives you a choice and then recommends that you take a particular choice, but doesn't require it. IMO sometimes this is done to get better acceptance at the cost of less-well-enforced freeness.
Slashdot Mirror
Could this be the cause of the Long-delayed echoes?
I understand that producing code by contract can be a way to make money. But unfortunately I can not cite such companies as successful, because they never release their damn financials! I have no proof for any company but Red Hat and MySQL.
Well, I have discussed this with several lawyers. In fact, I wrote it for the lawyers, and they paid me for it. Much of my work is being the bridge between engineering and legal departments, or the producer of Open Source policies for companies - which of course lawyers have to approve.
It sounds to me as if the license expects conveyance of source code along with the car, and you can't expect the average guy selling his car, or the average car dealer, to even know what source code is. And none of the lawyers have told me I'm full of it yet.
MySQL has worked out a successful means of being paid while producing Open Source. To do so they separate the people who want to share from those who do not, and get a fair quid-pro-quo from both. If proprietary software stopped existing this would not work, but for the forseeable future it does. Their proprietary licensed program is identical to the Open Source one (wasn't always, I know). As far as I can tell, it's all good.
Sale of an embedded device is distribution or conveyance of the software within the embedded device - even though it is not production of an additional copy. Here is the applicable GPL2 text regarding what you then must do if you distribute:
There is a third option in the GPL text that doesn't apply to commercial sale.
It really doesn't say you are responsible to provide the source of a modification. It says you have to provide the whole thing. It doesn't offer modification as a condition for this particular rule, just distribution.
The second and subsequent paragraphs of your argument are philosophical, and I think it's sufficient just to say I don't agree.
Thanks
Bruce
It's not use. It's creation of a derivative work. None of your scenario plays out until a derivative work is created.
I remember about 40 lawyers being in the room during some of the GPL3 meetings. GPL3 is the most legally-reviewed license we've seen, and the AGPL3 adds reasonably little text, which itself has had a pretty good review. I doubt it'll fall over so easily as you think.
I'd imagine that this particular vendor was influenced positively by what you'd call dogma. Holding a line is important. Even when it's not popular.
I don't think it will happen. If it does, we'll have to modify the license language.
I think we can indeed woo open hardware vendors, but I am much less confident that they can be the existing vendors. They just have too many conflicts of interest for it to work.
The first real piece of Open Hardware I've purchased is the ip08 PBX from David Rowe's store. No proprietary code at all in the box. It came with the screws that hold the box together and the rubber feet in a little zip-lock bag. In other words, the default mode of the unit when shipped is one in which you can "open the hood". No tricky process to crack the box open as with so many consumer products. And this even though the daughter cards came already installed. It's manufactured for David by Atcom, but the design is David's and is itself Open Sourced. Anyone can manufacture it.
It's using the Asterisk GUI 2.0, which is a little too fresh for the less technical, but it'll mature.
Typo. That's "HDMI". Anyway, nobody's fighting what is going on with HDMI.
Things are looking up a bit, but you noticed the big anti-net-neutrality putsch going on today, didn't you? And although audio seems to have opened up significantly, video is progressing in the opposite direction. The next iteration of HDML allows no analog video at all.
A more limited interpretation is that ownership of a copy embodies a limited right to use, period. Which doesn't bother us.
They said that about GPL 2 as well. Now it's getting difficult to buy some kinds of devices without GPL code in them.
My experience with the Sony DHG-HDD250 has been educational. I have two of these DVRs. They have the Busybox license (I created busybox) in the user manual. The user is not permitted to set the time. It can only be set by TV-Guide-On-Screen's signal. I guess that setting the time would allow circumvention of some DRM information somewhere. I suspect they planned to send DRM info over the TV-Guide-On-Screen signal. Or they do.
So, when the digital TV cut-over happened, SONY had not updated the unit to get the digital TV-Guide-On-Screen signal. So, the unit was effectively a brick. You couldn't set the time, and there was no program guide.
Sony finally did an update, some weeks after the cut-over. But they were under no obligation to do so, as by then all the units were out of warranty. I'd spent about $1000 to buy the pair and for a while they were useless.
Next time I'm buying open hardware. And I am licensing new code under GPL3 and AGPL3.
It originally came about because of the reverse-engineering of IBM BIOS. What has changed now is that devices are built to exercise strong DRM on digital media, and DMCA includes tremendous fines for circumvention of copyright-management systems, including the copyright of the device itself and its installed software.
The easiest way for us to be locked out would be if significant sites started to require TPM and a properly-signed OS for web access. They do want to protect their content from the evil "view source" button, etc.
I suspect that Wikipedia works that way, but I don't have proof. There are a lot of retired people with deep expertise that they would like to gift to the world, and the world is generally resistant to long lectures from old folks. So, the retirees put it in Wikipedia and feel fulfilled.
I don't see a reason for software development to transition to that demographic, though. You'd need some evidence that current retirees still want to program in significant and growing numbers - that they haven't already fulfilled that urge to their satisfaction. I think you are assuming that their enthusiasms won't change as they mature. And the numbers would have to be sufficient to overwhelm the presence of younger people involved in industry.
Yes, that is what dual-licensing is about. It allows you to make Open Source, and sell it as proprietary software to folks who don't want to free up their derivative works.
In both cases there is a quid-pro-quo. It is either software, or money.
We know of only two profitable Open Source companies, MySQL and Red Hat. The others have not released financials.
Richard wants the proprietary code to fail.
It can't mean asserting control, because the OSD does not require that any rights be withheld. Instead, it requires that rights beyond the default of "All Rights Reserved" must be conveyed. Public domain conveys all possible rights.
Well, BSD and Apache are also only Open Source if there is source code. Yet, the BSD and Apache licenses appear on the OSI web site as accepted licenses. The only reason "Public Domain" is not listed there is that it is not a license.
MySQL. They were profitable, and probably still are, and Sun bought them for 1.1 Billion dollars after they'd actually prepared, but not filed, an IPO. Too bad what happened to Sun after that, but it wasn't MySQL's fault.
FSF has that here. My own rationale is that application of DMCA to my code in a locked down device makes my code essentially proprietary. I might be persuaded to write proprietary code for someone, but I'd want to be paid for that. My motivation for writing code that I'm not paid for is that the code will remain free.
It's a lot like firewall code. You have to deal with a hostile environment.
For one of my customers I have defined processes that will allow the casual sale of a motor vehicle with Linux in the dashboard to happen without infringing the GPL. Technically, the seller should really give the buyer source code. But they aren't expected to be computer-literate. So, the manufacturer has to provide some services on their behalf.
Nobody is proposing to apply AGPL3 to Linux. But I am applying it to some new embedded software. And yes, if you perform it to clients over a network, the source code has to be made available.
It's Open Source if you have source code and the rights, stated in more detail in the OSD, to use, redistribute, and modify. Public domain would give you all those rights. The only remaining question once a program is public domain is "Is the source code available?" If it is, the program is really Open Source.
The TCP/IP was something called "winsock" from a third party, but it was definitely an add-on to 3.1 . As far as WfW, I think that out of the box there was the early SMB-based networking, not TCP/IP. And there was Novell.
You say that now, because you have motherboards available to you that don't even come with an operating system and are publicly documented so that you can run the OS you like. But your access to such things is by no means guaranteed. Consider a future in which embedded devices like phones are the way we use computers. Sometime in the future there could be no mass-market hardware that would allow Free Software to run. You won't be able to get the right code-signing, or a place in the app store. RMS was concerned about that. IMO for good reason. Thus the anti-circumvention material in GPL3.
By the way, we only get away with manufacturing hardware that runs Free Software for software-defined radio today because it's "test equipment". FCC has no intention of allowing open hardware for radio applications to reach the mass market. We could get into a similar situation with other open hardware.
Why doesn't he make LGPL into GPL? He gives you a choice and then recommends that you take a particular choice, but doesn't require it. IMO sometimes this is done to get better acceptance at the cost of less-well-enforced freeness.