First off that analogy I think proves the point. "beancounters" that is accountants don't ask employees to be in compliance with federal, state or SEC regs. Rather they ask for expense reports. Those expense reports are then reclassed in the ERP system, that generates P&L statements which go to finance. Finance modifies them further and that goes to the tax attorneys to be reclassed for the IRS. They don't require or expect employees to be in compliance with IRS policy. Rather accounting follows the business and isolates the company from IRS and SEC regulations to the point that you think expense reports are something useful in and of themselves.
Moreover, ask anyone in account about how much compliance they get on expense reports. Further IT in particular is notorious for charging activities to the wrong IONs (based on which projects have budget) and screwing up the depreciation schedules that finance creates.
No they shouldn't. IT should report into IT steering committees that are tasked with making sure IT meets corporate goals. Far too frequently IT departments don't fully understand the extra labor costs they are creating by not allowing technologies. IT departments don't get billed for lost productivity if they spend an extra 90 days doing a security study of a badly needed system. It might be cheaper net to do the security study in tandem with other parts of the project at 3, 4, 5x the cost to get the system out 75 days earlier.
Except that your job and your policies can interfere with their job. By your logic they can break your policies, because it is their job and it what they get paid to do, its not personal; and you should get over it and move along.
Or maybe you need to try and figure out what unmet business need is driving the desire for a new device and meet the need so they don't even want the new device.
IT policies and company policies can conflict and quite often do. Different people at different levels can issue different and conflict policies that employees have to work through.
(a) Boost sales 17% this quarter (b) Meet company X's security guidelines
Don't sound like they conflict. But what if company Y is the most likely sales target and Y needs flexible and nimble structures to support them?
There are natural conflicts because IT tradeoffs create losers and winners. Generally executives (and not just specifically IT executives) don't think through how to ameliorate the issues losers are going to have with the choices they made. That's the whole point of "getting to win-win" type strategies. People do have damn good reasons.
Oh I agree Munich is an upper bound in terms of cost / work. OTOH it is a an accurate task list. In the late 1990s and early 2000s many people had no clue how far their organizations had adopted a Windows culture and how complex this sort of undertaking would be. Even knowing what the upper bound looks like is a good thing for Linux at this point.
It depends what you mean by "contributing patches back". If they retain copyright and only grant you a GPL license then absolutely, those patches would lock the entire project. If they assign copyright or grant you an unlimited / unrestricted license then you can use them in a commercial version.
Assignment is fairly common for many large open source projects even when there are multiple contributors. Even if you don't intend to ever license under anything but one license creating ambiguous standing can be a major hinderance in enforcing copyright. For MIT there isn't much to enforce but for GPL it really matters. I have serious question as to how well small chunks of KDE or the Linux kernel could even be defended in court.
Yes Sun had a Windows culture. The extent of it was a shock to the executives at Sun, I believe. It certainly was an embarrassment that most of the employees at Sun were using Windows boxes to get their work done during a time when Sun was urging other companies to shift to Unix based solutions.
Scott McNealy might have been that anti-Windows but his people weren't.
Well first off lets not jump ahead. Microsoft's preliminary version of their store license agreement has some terms which indicate they may permit open source app to share without being entirely clear about mechanism. So we know that at least some people in Microsoft are thinking about that, that's different than Microsoft having fully worked through GPL issues.
Second, I'm not sure we fully understand the Microsoft security model. They may not be facing the same problems. Moreover if Microsoft does have the same problem and Apple and works through a solution Apple might adopt it. Apple, along with the FSF have rejected Tivo's theory about what is permitted under the GPLv2, and certainly the GPLv3 doesn't allow Tivo's approach. If Microsoft has a solution, one that they are willing to defend in court, Apple would have no reason not to fall in with their approach.
Third, the original claim was that Apple, "fights aggressively against free and open source software" which is a different criticism than "they don't spend time and effort creating an entire alternative mechanism to facilitate GPL software".
I understand the advantage of thin client, I'm a big proponent. I don't understand the purpose of cutting against the grain. If you want to go thin client use thin client apps and a thin client OS. If you have to go windows make the windows system's generic and you still get the provision and maintenance savings with thick client through something like View.
I never saw Citrix as a good solution for anything. Windows applications are designed to run on the local desktop. Windows is a thick client environment. If you want thin client, use a thin client OS and thin client applications. If you want Windows with easy maintenance something like VMWare View is fine.
Munich has been a financial disaster. There are no savings at this point. But it is not just OS licenses in real terms:
1) OS 2) Office 3) Sharepoint, SQL server... and all the associated collection of support servers 4) Advanced applications are all closed source (dynamics, universal communicator...)
That gets to around $2k / head up front and around $500 / yr. That's still not too bad as compared to maintaining a large IT department to do this all custom. In theory though most of this is generic and open source should be able to accommodate it.
In terms of large agencies that tried moving to Linux there were 3 main groups of companies
1) Companies that never had developed a Windows culture. Generally they were Unix shops (Sun, Sco primarily) and they were able to move to Linux easily.
2) Companies that were highly motivated tech companies: IBM, Oracle, Sun that all had a Windows culture. They had embarrassing failures in moving to Windows.
3) Companies that were not particularly technological and wanted to save money. The bag was mixed here but in general the costs got out of control and they threw in the towel.
Munich represents the one place where despite going way over time and budget they have kept plowing away. Demonstrating what it is actually going to take to move a large enterprise with a Windows culture over to Linux.
This is a BSD vs. GPL argument about which license better promotes code freedom. You are making a different argument which is about commercial vs. open source that end users should have the right to use closed source code that accomplishes their objectives. I think most people agree with you. I also think that is entirely irrelevant to the discussion of whether the BSD or GPL license better promotes code freedom.
Apple requires that someone submitting an app have either:
1) Copyright 2) A valid license which allows them to grant permission to Apple to redistribute per the app store mechanism.
You have to be legally entitled to grant Apple permission to redistribute. That is the law and has nothing to do with open source. Most people creating GPLed versions of apps don't have either (1) or (2). The FSF and Apple both agree the app store mechanism does not comply with the GPLv2 so people don't get (2) from the app being GPLed they need copyright or another license.
I'm not sure how Apple is fighting open source here. They are unable to comply with the terms of the license and thus are fully honoring the license by failing to redistribute. A copyright holder would be absolutely allowed to distribute a GPL app from the Apple Store.
That would be a nice theory except for the fact that open source was the norm in the 1950s and 60s. The internet, 1973, coincides with the collapse of open source as a norm. If you mean the mass use of internet for home, 1995/6, your problem is the modern open source movement takes off in 1994 with LAMP. The next phase of it takes place primarily in corporations. Open source for home never really takes off. Corporations had networks prior to 1995/6.
So no I don't think networking is the cause. As for BSD, BSD came out in 1977. BSD was part of the whole move of universities creating code which corporations put in their proprietary Unix products that were closed source. You can argue that BSD was part of the open systems movement, i.e. shared standards but I see no evidence it had anything to do with open source, open source was collapsing into irrelevance under BSD. Which BTW is precisely the GPL argument that the BSD license leads to widespread adoption of extended versions with closed copyright.
A writes some GPLv2 code call it X and puts it in the kernel. B takes X takes a subset of it X'. X' is under GPLv2. B takes X' and mixes it with some code Y for which he has copyright, creating X'+Y. He releases the the combined work under GPLv3. C takes X'+Y and violates the GPLv3 but not GPLv2. D takes X'+Y and violates both the GPLv3 and GPLv2.
1) Is B's release legal? Is GPLv2 compatible with GPLv3 or does the upgrade ban B from releasing X'+Y under GPLv3? Has B created a license which is unenforcible. If so, how is the new product licensed or is it just a violation of A's code and nothing more? Remember, Linus' claim is meaningless if it just applies to X', the real question is whether the can claim copyright authority over X'+Y on the basis of some sort of leadership over X'.
2) Can A sue B and win? Make it worse assume that X involves code from A1, A2, A3 and A4 with A1 and A2 granting permission while A3 and A4 don't. Does the fact that A1-A4 wrote their code solely for a collective work create collective responsibility? The popular answer on/. is no A3 and A4 retain all rights to subsets. But AFAIK courts do not look at things that way.
3) Can B sue C? Remember if the answer is yes then B has just upgraded the kernel to GPLv3. If the the answer is no, then you agree that once any part of the code license becomes questionable the whole thing unenforcible. This is exactly the situation with the entire kernel.
I know that it isn't included in the license, and that may or may not matter. It is in the explanatory text and it is in the form of the license (which is the FSF's version). I don't see how a kernel developer is going to win a lawsuit against someone distributing a GPLv3 kernel.
As for BSD the XNU kernel is not used by any other BSD. Free, Open and Net BSD have essentially nothing in common at a kernel level with OSX. Frankly the kernel that has the most Mach/XNU code in it other than OSX is NT which has some common code from a common (grand)parent, the Accent kernel. The contribution of Apple as far as the kernel is an open source XNU kernel which you could use in place of the BSD kernels to create a BSD.
The claim of BSD users is that it results in a situation that is more free. In other words it is being taken as a given that freedom for people downstream from the developers is an intrinsic good. If freedom is not an intrinsic good the whole anti-GPL argument made by the BSD camp, that the GPL reduces freedom, falls flat. Reducing freedom is no longer a bad thing.
So given that axiomatically freedom is an intrinsic good lets look at the X situation. At the time of the open systems movements there were essentially no important / useful Unixes with an open source derivative of X as their windowing system. The purpose of XFree86 was to create a free X first for Linux and BSD, they had to do this because there was none that was useful. Eventually (after about a decade) XFree86 became the reference implementation, and when this moved back to X.org the reference implementation became useful; but there was about 15 years where that wasn't the case. To this day many of the features of the proprietary X's are still not part of X.org.
Quite simply there was, in a meaningful sense no useful X that was open source. The reference implementations that existed, benefit operating system manufacturers but did not benefit the users of those operating systems. In effect there was no freedom, and since freedom is an intrinsic good that is a negative situation to be avoided.
"MIT licensed graphics library" = OpenGL which is a secondary library for Microsoft " or a BSD licensed network stack " = that when Microsoft first released their network stack there was a lot of BSD code in it.
1) Notice it starts in 2008. If had started in say 1995 you would see a huge drop in BSD/MIT... licenses to 70% GPL family in 2008. 2) I suspect this may have a lot to do with cell phone apps and the focus on web applications where the right to link freely and distribute might be more important.
There is some very serious questions of standing. Moreover, the GPL includes a clause allowing an upgrade. By releasing the code under a GPL license I think someone could release a version of the kernel under GPLv3 without developers agreeing. The controversy is that for the v3 clause to take effect Linus needs to invoke it and Linus doesn't believe he has that legal authority. ' As for Apple subsystems that are in BSD launchd was written in 2005 for BSD type systems ported as an option to FreeBSD that year launchd in FreeBSD 2008
Slashdot Mirror
First off that analogy I think proves the point. "beancounters" that is accountants don't ask employees to be in compliance with federal, state or SEC regs. Rather they ask for expense reports. Those expense reports are then reclassed in the ERP system, that generates P&L statements which go to finance. Finance modifies them further and that goes to the tax attorneys to be reclassed for the IRS. They don't require or expect employees to be in compliance with IRS policy. Rather accounting follows the business and isolates the company from IRS and SEC regulations to the point that you think expense reports are something useful in and of themselves.
Moreover, ask anyone in account about how much compliance they get on expense reports. Further IT in particular is notorious for charging activities to the wrong IONs (based on which projects have budget) and screwing up the depreciation schedules that finance creates.
No they shouldn't. IT should report into IT steering committees that are tasked with making sure IT meets corporate goals. Far too frequently IT departments don't fully understand the extra labor costs they are creating by not allowing technologies. IT departments don't get billed for lost productivity if they spend an extra 90 days doing a security study of a badly needed system. It might be cheaper net to do the security study in tandem with other parts of the project at 3, 4, 5x the cost to get the system out 75 days earlier.
Except that your job and your policies can interfere with their job. By your logic they can break your policies, because it is their job and it what they get paid to do, its not personal; and you should get over it and move along.
Or maybe you need to try and figure out what unmet business need is driving the desire for a new device and meet the need so they don't even want the new device.
IT policies and company policies can conflict and quite often do. Different people at different levels can issue different and conflict policies that employees have to work through.
(a) Boost sales 17% this quarter
(b) Meet company X's security guidelines
Don't sound like they conflict. But what if company Y is the most likely sales target and Y needs flexible and nimble structures to support them?
There are natural conflicts because IT tradeoffs create losers and winners. Generally executives (and not just specifically IT executives) don't think through how to ameliorate the issues losers are going to have with the choices they made. That's the whole point of "getting to win-win" type strategies. People do have damn good reasons.
Oh I agree Munich is an upper bound in terms of cost / work. OTOH it is a an accurate task list. In the late 1990s and early 2000s many people had no clue how far their organizations had adopted a Windows culture and how complex this sort of undertaking would be. Even knowing what the upper bound looks like is a good thing for Linux at this point.
It depends what you mean by "contributing patches back". If they retain copyright and only grant you a GPL license then absolutely, those patches would lock the entire project. If they assign copyright or grant you an unlimited / unrestricted license then you can use them in a commercial version.
Assignment is fairly common for many large open source projects even when there are multiple contributors. Even if you don't intend to ever license under anything but one license creating ambiguous standing can be a major hinderance in enforcing copyright. For MIT there isn't much to enforce but for GPL it really matters. I have serious question as to how well small chunks of KDE or the Linux kernel could even be defended in court.
Yes Sun had a Windows culture. The extent of it was a shock to the executives at Sun, I believe. It certainly was an embarrassment that most of the employees at Sun were using Windows boxes to get their work done during a time when Sun was urging other companies to shift to Unix based solutions.
Scott McNealy might have been that anti-Windows but his people weren't.
Well first off lets not jump ahead. Microsoft's preliminary version of their store license agreement has some terms which indicate they may permit open source app to share without being entirely clear about mechanism. So we know that at least some people in Microsoft are thinking about that, that's different than Microsoft having fully worked through GPL issues.
Second, I'm not sure we fully understand the Microsoft security model. They may not be facing the same problems. Moreover if Microsoft does have the same problem and Apple and works through a solution Apple might adopt it. Apple, along with the FSF have rejected Tivo's theory about what is permitted under the GPLv2, and certainly the GPLv3 doesn't allow Tivo's approach. If Microsoft has a solution, one that they are willing to defend in court, Apple would have no reason not to fall in with their approach.
Third, the original claim was that Apple, "fights aggressively against free and open source software" which is a different criticism than "they don't spend time and effort creating an entire alternative mechanism to facilitate GPL software".
I understand the advantage of thin client, I'm a big proponent. I don't understand the purpose of cutting against the grain. If you want to go thin client use thin client apps and a thin client OS. If you have to go windows make the windows system's generic and you still get the provision and maintenance savings with thick client through something like View.
This is written sarcastically but there is some actual content I didn't know.
I'm not sure what you are trying to say with the client side decorations list. Can you link to a discussion?
As for how copy/paste should work. Hopefully Linux gets a full featured OLE, type copy paste finally.
I never saw Citrix as a good solution for anything. Windows applications are designed to run on the local desktop. Windows is a thick client environment. If you want thin client, use a thin client OS and thin client applications. If you want Windows with easy maintenance something like VMWare View is fine.
Munich has been a financial disaster. There are no savings at this point. But it is not just OS licenses in real terms:
1) OS
2) Office
3) Sharepoint, SQL server... and all the associated collection of support servers
4) Advanced applications are all closed source (dynamics, universal communicator...)
That gets to around $2k / head up front and around $500 / yr. That's still not too bad as compared to maintaining a large IT department to do this all custom. In theory though most of this is generic and open source should be able to accommodate it.
Just info for younger people on /.
In terms of large agencies that tried moving to Linux there were 3 main groups of companies
1) Companies that never had developed a Windows culture. Generally they were Unix shops (Sun, Sco primarily) and they were able to move to Linux easily.
2) Companies that were highly motivated tech companies: IBM, Oracle, Sun that all had a Windows culture. They had embarrassing failures in moving to Windows.
3) Companies that were not particularly technological and wanted to save money. The bag was mixed here but in general the costs got out of control and they threw in the towel.
Munich represents the one place where despite going way over time and budget they have kept plowing away. Demonstrating what it is actually going to take to move a large enterprise with a Windows culture over to Linux.
This is a BSD vs. GPL argument about which license better promotes code freedom. You are making a different argument which is about commercial vs. open source that end users should have the right to use closed source code that accomplishes their objectives. I think most people agree with you. I also think that is entirely irrelevant to the discussion of whether the BSD or GPL license better promotes code freedom.
Apple requires that someone submitting an app have either:
1) Copyright
2) A valid license which allows them to grant permission to Apple to redistribute per the app store mechanism.
You have to be legally entitled to grant Apple permission to redistribute. That is the law and has nothing to do with open source. Most people creating GPLed versions of apps don't have either (1) or (2). The FSF and Apple both agree the app store mechanism does not comply with the GPLv2 so people don't get (2) from the app being GPLed they need copyright or another license.
I'm not sure how Apple is fighting open source here. They are unable to comply with the terms of the license and thus are fully honoring the license by failing to redistribute. A copyright holder would be absolutely allowed to distribute a GPL app from the Apple Store.
You are right. OK so statutory minimums aren't a problem.
That would be a nice theory except for the fact that open source was the norm in the 1950s and 60s. The internet, 1973, coincides with the collapse of open source as a norm. If you mean the mass use of internet for home, 1995/6, your problem is the modern open source movement takes off in 1994 with LAMP. The next phase of it takes place primarily in corporations. Open source for home never really takes off. Corporations had networks prior to 1995/6.
So no I don't think networking is the cause. As for BSD, BSD came out in 1977. BSD was part of the whole move of universities creating code which corporations put in their proprietary Unix products that were closed source. You can argue that BSD was part of the open systems movement, i.e. shared standards but I see no evidence it had anything to do with open source, open source was collapsing into irrelevance under BSD. Which BTW is precisely the GPL argument that the BSD license leads to widespread adoption of extended versions with closed copyright.
That's not quite the issue.
A writes some GPLv2 code call it X and puts it in the kernel.
B takes X takes a subset of it X'. X' is under GPLv2.
B takes X' and mixes it with some code Y for which he has copyright, creating X'+Y. He releases the the combined work under GPLv3.
C takes X'+Y and violates the GPLv3 but not GPLv2.
D takes X'+Y and violates both the GPLv3 and GPLv2.
1) Is B's release legal? Is GPLv2 compatible with GPLv3 or does the upgrade ban B from releasing X'+Y under GPLv3? Has B created a license which is unenforcible. If so, how is the new product licensed or is it just a violation of A's code and nothing more? Remember, Linus' claim is meaningless if it just applies to X', the real question is whether the can claim copyright authority over X'+Y on the basis of some sort of leadership over X'.
2) Can A sue B and win? Make it worse assume that X involves code from A1, A2, A3 and A4 with A1 and A2 granting permission while A3 and A4 don't. Does the fact that A1-A4 wrote their code solely for a collective work create collective responsibility? The popular answer on /. is no A3 and A4 retain all rights to subsets. But AFAIK courts do not look at things that way.
3) Can B sue C? Remember if the answer is yes then B has just upgraded the kernel to GPLv3. If the the answer is no, then you agree that once any part of the code license becomes questionable the whole thing unenforcible. This is exactly the situation with the entire kernel.
4) Can B sue D? What about A can he sue D?
I don't think this is clear cut at all.
I know that it isn't included in the license, and that may or may not matter. It is in the explanatory text and it is in the form of the license (which is the FSF's version). I don't see how a kernel developer is going to win a lawsuit against someone distributing a GPLv3 kernel.
As for BSD the XNU kernel is not used by any other BSD. Free, Open and Net BSD have essentially nothing in common at a kernel level with OSX. Frankly the kernel that has the most Mach/XNU code in it other than OSX is NT which has some common code from a common (grand)parent, the Accent kernel. The contribution of Apple as far as the kernel is an open source XNU kernel which you could use in place of the BSD kernels to create a BSD.
The claim of BSD users is that it results in a situation that is more free. In other words it is being taken as a given that freedom for people downstream from the developers is an intrinsic good. If freedom is not an intrinsic good the whole anti-GPL argument made by the BSD camp, that the GPL reduces freedom, falls flat. Reducing freedom is no longer a bad thing.
So given that axiomatically freedom is an intrinsic good lets look at the X situation. At the time of the open systems movements there were essentially no important / useful Unixes with an open source derivative of X as their windowing system. The purpose of XFree86 was to create a free X first for Linux and BSD, they had to do this because there was none that was useful. Eventually (after about a decade) XFree86 became the reference implementation, and when this moved back to X.org the reference implementation became useful; but there was about 15 years where that wasn't the case. To this day many of the features of the proprietary X's are still not part of X.org.
Quite simply there was, in a meaningful sense no useful X that was open source. The reference implementations that existed, benefit operating system manufacturers but did not benefit the users of those operating systems. In effect there was no freedom, and since freedom is an intrinsic good that is a negative situation to be avoided.
I think he is talking about Microsoft.
"MIT licensed graphics library" = OpenGL which is a secondary library for Microsoft
" or a BSD licensed network stack " = that when Microsoft first released their network stack there was a lot of BSD code in it.
I'd like to comment on the graph.
1) Notice it starts in 2008. If had started in say 1995 you would see a huge drop in BSD/MIT... licenses to 70% GPL family in 2008.
2) I suspect this may have a lot to do with cell phone apps and the focus on web applications where the right to link freely and distribute might be more important.
This was a pretty good detailed comment. You might want to consider getting an account.
There is some very serious questions of standing. Moreover, the GPL includes a clause allowing an upgrade. By releasing the code under a GPL license I think someone could release a version of the kernel under GPLv3 without developers agreeing. The controversy is that for the v3 clause to take effect Linus needs to invoke it and Linus doesn't believe he has that legal authority.
'
As for Apple subsystems that are in BSD
launchd was written in 2005 for BSD type systems
ported as an option to FreeBSD that year
launchd in FreeBSD 2008