You should keep the copyright there because the copyright is there. That's the law. If you do not want to keep the copyright, you may rewrite the file. There are many things that would change: identation, variable names, naming conventions (use of underscore, upper/lower case, noun/verb order), structures that are used by the _driver_, instead of being things that a read/written from the hardware, placement of comments, style of comments, number of comments, content of comments, wording of comments, order of definitions in the file...
Just because the header file is used for some kind of hardware doesn't mean, _at all_, that they will look alike. Why don't you look at header files for the same device that do not share code between Linux and FreeBSD? As a matter of fact, why don't you look at the same header file on different versions or different drivers for the same hardware on the same operating system?
'Apple simply found a source
of cheap high-quality systems software that it could make its
own without needing to give back so much as a bug fix, let
alone useful software projects.'
Well, that proves the BSD license does what it set out to do: make high-quality code widely used, thus setting a higher standard for all.
Not at all. It's you that are being pig-headed. I was clear enough: the kerberos problem with Microsoft was a result of a problem with the protocol, not a problem with the license.
Microsoft embraced and extended the protocol,
Correct. More to the point, they used a loophole in the protocol so what they wrote is technically Kerberos but as a matter of fact result in incompatibilities between computers running their implementation and computers running others.
and almost definitely the code (why not,
when it's there for the taking?).
So you think they used the code, eh? Well, FYI, they didn't.
Thanks to their proprietary business model,
Well, a true sentence. You are improving.
aided and
abetted by freely available code licensed under BSD-like licenses,
Huh? In what way? Explain. And, more to the point, explain exactly what would have happened differently if Kerberos had only been available as GPL code. Remember: they wrote their own implementation.
free software
currently has no easy legal way to be to compatible.
Protocol design. Either you forbid any extensions, or people can screw up. And they can screw up even if extensions are forbidden. As a matter of fact, they can screw up even without extending anything (hello 3Com Home Connect Dual Speed!).
On the other hand, they could use a proprietary, non-documented protocol, which they could extend to keep things incompatible as they see fit (hello AOL!), or even patent the whole protocol. See real world examples.
You must understand that I am taking a long term view. In the long term, producing code
under a BSD-style license may encourage vendors to adopt 'standards' but it allows them to
extend and appropriate them too.
Corporations can extend any standard, no matter the license of existing source code implementing it. The license is utterly irrelevant.
In other words, it encourages the Microsoft business
model.
Yeah, you are right. If they simply did not adopt the standard at all, they wouldn't extend/appropriate it.
Of course, unless the standard became popular. In which case they would simply write their own version of whatever it is and extend/appropriate it the same way, and then kill off the competition with marketing and market share. Just like they do nowadays.
Code released under viral Free-software licenses is different. True, traditional software
houses won't base products upon it initially.
s/initially/ever/
Unless it's a stand-alone product, in which case the virus is irrelevant.
But if enough good viral-free software is
produced and used, then businesses lose the option. They either compete against the
software head on, or build upon it and release more viral-free software.
Go on...
If they compete
against it, they are at a disadvantage compared to those who work with it, because they
can't build on the freely available work.
Yeah, right, like anyone does.
Code reusability is a myth. Libraries, yes. Simple-minded routines here and there, skeletons for things like drivers. Beyond that, which is no real advantage, code isn't reused.
But the real problem with this line of thought is pretty simple: it assumes the best code wins. When did that ever happen? Get real.
If they work with it, they can't appropriate it, and
the user wins. Code released under BSD-style licenses just slows this process down.
Excellent analysis. Of course, the premises are flawed, but if they were true, then you would be right.
The prevalence of TCP/IP has less to do with no-strings free implementations and more to
do with a large pre-existing infrastructure developed by the defence and research
communities, and that it works very well, if we ignore security.
Suuuuuuuuuuuuuure. I wonder how did we run Bitnet, the research community network, so long without a TCP/IP stack. And I look at all the other very extensive networks and wonder what happened to them. I look at the world-wide X.25 and wonder where did it go.
And I look back and all I can recall is how the only wan protocol I could run on my PC/XT was TCP/IP.
There's really only one thing FreeBSD offered before other OS (maybe Debian already have developed an alternative, though) that would be useful to you (as far as I know), jail.
On the other hand, both Linux and FreeBSD have been able to deal with the problem you describe like forever. Just set limits on login.conf for FreeBSD, and probably something similar on Debian.
specifically, viral
Free software, like GPL'ed softare, benefits users, because viral Free software promotes
real open standards, i.e. freedom from proprietary 'intellectual property'. The BSD
license and it's like do not seem to do this, witness Kerberos amongst others.
You seem misinformed. The Kerberos thing was a problem with the protocol specification, and had absolutely nothing to do with any software license whatsoever.
On the other hand, BSD code promotes standards because vendors actually use them. You _won't_ see Microsoft use GPL code in their mainstream products. Ever. Period.
That's why TPC/IP is everywhere, for example. There were many networking protocols in the 80s, but only one which had a free implementation available without strings so every OS maker under the sun added it to their operating systems.
We have to things to thank for for the Release Notes.
First, we use a source management tool (namely CVS), so all changes made to the source code are documented at the time they are made. This makes it much easier to keep track of what they are.
Second, Bruce A. Mah, who volunteered for the generally tedious but very important job of actually reading the commit logs and then keeping the Release Notes up to date for both -current and -stable branches.
I think she's on the right track, but most of the
places she says "Linux" I would substitute "one distribution". GNU/Linux need not
be a monolithic entity to be adopted, there just has to be one user-proof distro
available.
Sheesh, didn't anything in the article actually get inside your head? The very existence of multiple distributions is a problem.
I failed to mention many things about the series. Book titles, for instance. I'm not reviewing it.:-)
And as for the fleet, it's not known (at Startide Rising) whether it's really progenitor's or not (though it is certainly suspected so), and there are many species which are humanoid.
It's funny how so many people think the national debt is a bad thing, when withing reason (e.g. the runup during
Reagan's years rather than that of Bush Sr.), it's part of the system that has been so successful.
There is a big difference between what the people do and what the government do. National debt has a huge negative impact on the economy, unless someone is financing the national currency (like the world's investors finance the dollar).
Well, of course the Canadians waste far more water than the americans. They have the world's largest water resources and a relatively small population.
This expression comes from the Uplift series, a couple of science fiction trilogies by David Brin. The central theme in the series is that no sentient species in the universe, except one fabled "progenitors", has ever attained sentience by itself. Instead, they all have been genetically engineered into sentience, a process known as "uplifting".
Humans, naturally, uplift dolphins and chimpanzees (and then proceed to dogs, what a waste!).
Mind you, I hated the first book, the second book is terribly annoying and it's story is continued (and finished) only in the fifth and sixth books. Third book is actually ok, but no dolphins there.:-)
Yes, that's one possible scenario. You haven't shown how it is the only possible scenario, nor even why it even
the most likely scenario.
Value-added features. Everyone had an operating system. Features exclusive to one's own OS were (hell, still are) selling points, advantages one had over the competition used to entice customers.
Opening the source code would negate that advantage, which was considered very important (lately it has been downgraded to just important). It would be unthinkable for them to do that, and if you believe otherwise I have this nice open bridge I'd like to sell you.
"everyone" didn't think that ISO/OSI was the Next Big Thing, certainly not at the time that TCP/IP was
introduced and became popular.
Oh, no? Perhaps it was just my impression that OSI was in the the plans of every telco in the world, that OSI was the most talked-about networking protocol in the specialized magazines, that the very study of network protocols was centered around OSI, that every big OS company and telco were part of the OSI design committee?
Or perhaps my memory of that time is not as good as yours. Who knows...
I'm not sure how you can say that TCP/IP wouldn't have taken off if it had first been made available under the
GPL. It's quite possible that we'd not only still be using TCP/IP, but that a whole lot more software would now
be available under the GPL instead of being proprietary with fractured standards.
Well, it would work like this. One day, a meeting is called with Company X top developers to discuss the introduction of a networking stack on their operating system. They have really very few choices. They can develop their own protocol from scratch, but that would be costly and would take them very long, which would put them at a disadvantage against the competition. They could buy the network stack from someone else, but that's playing into competition's hand, and they would always stay at a disadvantage, so that's not really an option. And they can write a network stack for a public protocol. Of course, all being educated people, they know the Big Thing around the corner is ISO/OSI (I MEAN it -- that's what everyone thought until early 90s). At this point, though, one developer makes the following remark:
"Hey, some researchers in University X created a network protocol and released the source code free of cost. All we would have to do is release the source code for our whole operating system if we were to use their code!"
At this point the whole room falls into histerical laughter and, having broken the tension, they set forth to write their OSI stack.
When I said "TCP/IP hadn't been available under the BSD license", I meant an implementation of said protocols was available under the BSD license.
As a result, the various OS makers had two options:
1) Develop an OSI implementation from scratch.
2) Use an implementation based on the freely available (and modifiable/redistributable, even under closed source licenses) BSD TCP/IP stack.
That is why Internet actually works. That's why an array so huge of different operating systems could actually connected with each other without compatibility problems. I mean, we are talking about people who couldn't settle for a common floppy disk format here.
Anyway, if this people *didn't* have access to a TCP/IP implementation they could do whatever they wanted with, they would end up implementing the bloated and accounting-oriented OSI. Notice that OSI is an actual international standard, while TCP/IP is not (just a de facto one).
Perhaps there is some confusion here. If I release a software under BSDL, for example, that software will *STAY* under that license, period. I can release it under other licenses, as the copyright-owner, but that does not "remove" that source code with BSD license on it from existence.
The question is whether we derivative works of that software to be kept open, or if we want that software to be taken maximum advantage off.
Since we would all be using OSI if TCP/IP hadn't been available under the BSD license, I know to which camp I belong.
Slashdot Mirror
Do you write source code?
If you do, please tell me what you do. I'll start copying your code and leaving your name by oversight.
You should keep the copyright there because the copyright is there. That's the law. If you do not want to keep the copyright, you may rewrite the file. There are many things that would change: identation, variable names, naming conventions (use of underscore, upper/lower case, noun/verb order), structures that are used by the _driver_, instead of being things that a read/written from the hardware, placement of comments, style of comments, number of comments, content of comments, wording of comments, order of definitions in the file...
Just because the header file is used for some kind of hardware doesn't mean, _at all_, that they will look alike. Why don't you look at header files for the same device that do not share code between Linux and FreeBSD? As a matter of fact, why don't you look at the same header file on different versions or different drivers for the same hardware on the same operating system?
It was copied verbatim, minus the copyright notice.
It matters because it is always useful to know who steals code without due credit.
The copyright notice must appear on binary-only distributions too.
'Apple simply found a source
of cheap high-quality systems software that it could make its
own without needing to give back so much as a bug fix, let
alone useful software projects.'
Well, that proves the BSD license does what it set out to do: make high-quality code widely used, thus setting a higher standard for all.
...and, of course, snapshots are also supported by FreeBSD.
Not at all. It's you that are being pig-headed. I was clear enough: the kerberos problem with Microsoft was a result of a problem with the protocol, not a problem with the license.
Microsoft embraced and extended the protocol,
Correct. More to the point, they used a loophole in the protocol so what they wrote is technically Kerberos but as a matter of fact result in incompatibilities between computers running their implementation and computers running others.
and almost definitely the code (why not, when it's there for the taking?).
So you think they used the code, eh? Well, FYI, they didn't.
Thanks to their proprietary business model,
Well, a true sentence. You are improving.
aided and abetted by freely available code licensed under BSD-like licenses,
Huh? In what way? Explain. And, more to the point, explain exactly what would have happened differently if Kerberos had only been available as GPL code. Remember: they wrote their own implementation.
free software currently has no easy legal way to be to compatible.
Protocol design. Either you forbid any extensions, or people can screw up. And they can screw up even if extensions are forbidden. As a matter of fact, they can screw up even without extending anything (hello 3Com Home Connect Dual Speed!).
On the other hand, they could use a proprietary, non-documented protocol, which they could extend to keep things incompatible as they see fit (hello AOL!), or even patent the whole protocol. See real world examples.
You must understand that I am taking a long term view. In the long term, producing code under a BSD-style license may encourage vendors to adopt 'standards' but it allows them to extend and appropriate them too.
Corporations can extend any standard, no matter the license of existing source code implementing it. The license is utterly irrelevant.
In other words, it encourages the Microsoft business model.
Yeah, you are right. If they simply did not adopt the standard at all, they wouldn't extend/appropriate it.
Of course, unless the standard became popular. In which case they would simply write their own version of whatever it is and extend/appropriate it the same way, and then kill off the competition with marketing and market share. Just like they do nowadays.
Code released under viral Free-software licenses is different. True, traditional software houses won't base products upon it initially.
s/initially/ever/
Unless it's a stand-alone product, in which case the virus is irrelevant.
But if enough good viral-free software is produced and used, then businesses lose the option. They either compete against the software head on, or build upon it and release more viral-free software.
Go on...
If they compete against it, they are at a disadvantage compared to those who work with it, because they can't build on the freely available work.
Yeah, right, like anyone does.
Code reusability is a myth. Libraries, yes. Simple-minded routines here and there, skeletons for things like drivers. Beyond that, which is no real advantage, code isn't reused.
But the real problem with this line of thought is pretty simple: it assumes the best code wins. When did that ever happen? Get real.
If they work with it, they can't appropriate it, and the user wins. Code released under BSD-style licenses just slows this process down.
Excellent analysis. Of course, the premises are flawed, but if they were true, then you would be right.
The prevalence of TCP/IP has less to do with no-strings free implementations and more to do with a large pre-existing infrastructure developed by the defence and research communities, and that it works very well, if we ignore security.
Suuuuuuuuuuuuuure. I wonder how did we run Bitnet, the research community network, so long without a TCP/IP stack. And I look at all the other very extensive networks and wonder what happened to them. I look at the world-wide X.25 and wonder where did it go.
And I look back and all I can recall is how the only wan protocol I could run on my PC/XT was TCP/IP.
Funny that while you criticize FreeBSD's NFS, you seem to have forgotten that Linux NFS is broken beyond hope.
Check this man page, and the links from it.
On the other hand, both Linux and FreeBSD have been able to deal with the problem you describe like forever. Just set limits on login.conf for FreeBSD, and probably something similar on Debian.
You seem misinformed. The Kerberos thing was a problem with the protocol specification, and had absolutely nothing to do with any software license whatsoever.
On the other hand, BSD code promotes standards because vendors actually use them. You _won't_ see Microsoft use GPL code in their mainstream products. Ever. Period.
That's why TPC/IP is everywhere, for example. There were many networking protocols in the 80s, but only one which had a free implementation available without strings so every OS maker under the sun added it to their operating systems.
We have to things to thank for for the Release Notes.
First, we use a source management tool (namely CVS), so all changes made to the source code are documented at the time they are made. This makes it much easier to keep track of what they are.
Second, Bruce A. Mah, who volunteered for the generally tedious but very important job of actually reading the commit logs and then keeping the Release Notes up to date for both -current and -stable branches.
Let me take this opportunity to thank him for it.
Ok, I can see the potential for popcorn-making, but... while I like butter on my popcorn, petroleum somehow is just not the same thing.
ps: I loved the quote. :-)
Sheesh, didn't anything in the article actually get inside your head? The very existence of multiple distributions is a problem.
No, I'm not kidding.
Speculative Fiction? Things like the Trojan Horse Project??? Well, that most definitely explains it. I *hate* the genre.
Seriously, I never heard of him before. What did he write for anyone to give a damn about what he says?
I failed to mention many things about the series. Book titles, for instance. I'm not reviewing it. :-)
And as for the fleet, it's not known (at Startide Rising) whether it's really progenitor's or not (though it is certainly suspected so), and there are many species which are humanoid.
There is a big difference between what the people do and what the government do. National debt has a huge negative impact on the economy, unless someone is financing the national currency (like the world's investors finance the dollar).
Well, of course the Canadians waste far more water than the americans. They have the world's largest water resources and a relatively small population.
The US produce more garbage and more global warming gases on a per capita basis than any other country in the world by a LARGE margin.
If the rest of the world produced the same amount of pollutants as the US on a per capita basis... well, there would be no world left.
This expression comes from the Uplift series, a couple of science fiction trilogies by David Brin. The central theme in the series is that no sentient species in the universe, except one fabled "progenitors", has ever attained sentience by itself. Instead, they all have been genetically engineered into sentience, a process known as "uplifting".
:-)
Humans, naturally, uplift dolphins and chimpanzees (and then proceed to dogs, what a waste!).
Mind you, I hated the first book, the second book is terribly annoying and it's story is continued (and finished) only in the fifth and sixth books. Third book is actually ok, but no dolphins there.
Value-added features. Everyone had an operating system. Features exclusive to one's own OS were (hell, still are) selling points, advantages one had over the competition used to entice customers.
Opening the source code would negate that advantage, which was considered very important (lately it has been downgraded to just important). It would be unthinkable for them to do that, and if you believe otherwise I have this nice open bridge I'd like to sell you.
"everyone" didn't think that ISO/OSI was the Next Big Thing, certainly not at the time that TCP/IP was introduced and became popular.
Oh, no? Perhaps it was just my impression that OSI was in the the plans of every telco in the world, that OSI was the most talked-about networking protocol in the specialized magazines, that the very study of network protocols was centered around OSI, that every big OS company and telco were part of the OSI design committee?
Or perhaps my memory of that time is not as good as yours. Who knows...
Well, it would work like this. One day, a meeting is called with Company X top developers to discuss the introduction of a networking stack on their operating system. They have really very few choices. They can develop their own protocol from scratch, but that would be costly and would take them very long, which would put them at a disadvantage against the competition. They could buy the network stack from someone else, but that's playing into competition's hand, and they would always stay at a disadvantage, so that's not really an option. And they can write a network stack for a public protocol. Of course, all being educated people, they know the Big Thing around the corner is ISO/OSI (I MEAN it -- that's what everyone thought until early 90s). At this point, though, one developer makes the following remark:
"Hey, some researchers in University X created a network protocol and released the source code free of cost. All we would have to do is release the source code for our whole operating system if we were to use their code!"
At this point the whole room falls into histerical laughter and, having broken the tension, they set forth to write their OSI stack.
When I said "TCP/IP hadn't been available under the BSD license", I meant an implementation of said protocols was available under the BSD license.
As a result, the various OS makers had two options:
1) Develop an OSI implementation from scratch.
2) Use an implementation based on the freely available (and modifiable/redistributable, even under closed source licenses) BSD TCP/IP stack.
That is why Internet actually works. That's why an array so huge of different operating systems could actually connected with each other without compatibility problems. I mean, we are talking about people who couldn't settle for a common floppy disk format here.
Anyway, if this people *didn't* have access to a TCP/IP implementation they could do whatever they wanted with, they would end up implementing the bloated and accounting-oriented OSI. Notice that OSI is an actual international standard, while TCP/IP is not (just a de facto one).
Perhaps there is some confusion here. If I release a software under BSDL, for example, that software will *STAY* under that license, period. I can release it under other licenses, as the copyright-owner, but that does not "remove" that source code with BSD license on it from existence.
The question is whether we derivative works of that software to be kept open, or if we want that software to be taken maximum advantage off.
Since we would all be using OSI if TCP/IP hadn't been available under the BSD license, I know to which camp I belong.