Matthew Garrett: How-To Drive Developers From OS X To Linux

[Luke_Wolf]

On this we are in full agreement. I guess we also agree that the world is not a perfect place, and that human beings need some kind of common understanding to collaborate effectively. Anarchy has been tried and tested on numerous occasions, it typically exposes human beings ability of cruelty and self destruction.

Actually there is an occurrence where both anarchy and communalism (as well as dictatorships, republics, and all sorts of other systems) have been tried to great success, which is to say the OSS community. The difference between OSS and all other systems is that there is no cost to consume software. As a result no matter how many users will never contribute back to a project their effect on the software is 0. As a result while resources (such as the number of software developers) are still finite, models that typically fall apart due to limitations of resources and dead weight no longer have such issues. It is the first and likely only time a utopia will ever show up in human existence.

This underlines why I believe copy-left is so successful. I believe it is all about the common understanding that the copy-left project is about making an open source code through a collaborative model. Copyright transfer agreements (with the possible exception of the fsf) tends to destroy this though. Permissive licensing sends out a very different signal to all developers. It signals that the intended usage of contributions is as part of a larger proprietary software.

Actually I would argue that copy-left and permissive software send out the same signal, however the difference between the two is the difference between a proverbial stick and a proverbial carrot. Those who support copy-left basically believe that they have to use the stick to force other developers to be a part of the community, whereas those who support permissive licenses believe that the open source methodology wins on it's own merits and that by opening up the target audience to anyone and everyone to do whatever they want with it (including licensing it under a proprietary license) that more people are likely to become involved in the community by offering them the carrot (which in Apache's case includes moving to kill off software patents which nobody who actually cares about software wanted in the first place).

[Hamish Wilson]

Actually I would argue that copy-left and permissive software send out the same signal, however the difference between the two is the difference between a proverbial stick and a proverbial carrot. Those who support copy-left basically believe that they have to use the stick to force other developers to be a part of the community, whereas those who support permissive licenses believe that the open source methodology wins on it's own merits and that by opening up the target audience to anyone and everyone to do whatever they want with it (including licensing it under a proprietary license) that more people are likely to become involved in the community by offering them the carrot (which in Apache's case includes moving to kill off software patents which nobody who actually cares about software wanted in the first place).

The problem with this train of thought, and a lot of what Eric S. Raymond tried to claim over a decade ago, is that Open Source is not superior in every instance compared to other development methods. Let us not forget that while "free software" has been considered a muddled term due to its connotation in the English language with the foreign word "gratis", the term "Open Source" is also just as muddled and in some ways arguably more confusing due to it being applied both to software under a "free" or "open" license as well as being applied to projects which are developed under an "open" development model. While some projects that use an open development model do flourish, such as the Linux kernel, Apache, and other such projects, it has also proved itself to be terrible for certain others areas where a more tight control would be of benefit, such as game development. There are some examples were this does work, such as with Battle for Wesnoth which found a way to take thousands of different contributions and make it into a coherent whole, but in general it is far better to have a small closed development team for games, whether they in the end be released proprietary or under a free software license.

Another area from gaming which is actually in support of copyleft from the corporate perspective is the release of video game engines, such what John Carmack has done. Carmack choose the GNU GPL in order to keep development on these engines open but at the same time limit its effect on their engine licensing business. This way the community could do its magic to improve the companies older games while at the same time forcing people who wanted to use the code for their own proprietary game projects to still need to get a license from id Software.

The bottom line here is that neither copy-left or permissive is really any more "pragmatic" than the other, as the most practical choose is not always the same one in each and every instance.

[Del_]

Actually there is an occurrence where both anarchy and communalism (as well as dictatorships, republiveryrts of other systems) have been tried to great success, which is to say the OSS community.

No, either you are unaware or you choose to ignore realities. Public domain code failed utterly (yes it was quite popular). I could argue on the merits of permissive licensing too, but the conclusions are less clear cut.

The difference between OSS and all other systems is that there is no cost to consume software.

Yes, but this also goes for the cost of copying and distributing software. What this means is that there are very strong forces favouring the market leader, resulting in dominating player within each software category. This is why permissive licensing fails in terms of collaborative company effort. There will only be one winner when it comes to open stack, unless it goes copy-left. Just as any company putting resources into llvm/clang is pouring resources into Apple, strengthening a competitor who may put them out of business (yes, I am looking at you Intel with your compiler suite).

Actually I would argue that copy-left and permissive software send out the same signal.

Feel free to argue all you want, but you are wrong. It is not the same signal simply because it really isn't the same signal. That is why Theo De Radt needs to be very vocal about what he wants to achieve. If you slam gpl on your code, we all get the message. You cannot argue your way out of reality.

[FatBSDLoser]

The failure of GCC and the rise and domination of LLVM/Clang is an amazing example of what happens when developers are liberated from poisonous viral licences used to promote nutty ideologies like the GPL.

Development, code sharing, and technological advances in the compiler world have exploded thanks to LLVM/Clang having a truly free and open license.

Wrong, it is the failure of Clang and the continual dominance of GCC that is an amazing example of what happens when developers, contributors and users are liberated from poisonous proprietary and pro-lockup licenses like BSD, MIT and CDDL used to promote retarded self-harming practices like BSD, Solaris and Minix.

It is the GNU Compiler Collection (GCC) and GNU binutils being GPL that caused the explosion of development, code sharing and technological advances in the compiler world in which CLANG is desperately trying to feed off and give nothing back like the parasite it really is.

[TeamBlackFox]

Wrong, it is the failure of Clang and the continual dominance of GCC that is an amazing example of what happens when developers, contributors and users are liberated from poisonous proprietary and pro-lockup licenses like BSD, MIT and CDDL used to promote retarded self-harming practices like BSD, Solaris and Minix.

It is the GNU Compiler Collection (GCC) and GNU binutils being GPL that caused the explosion of development, code sharing and technological advances in the compiler world in which CLANG is desperately trying to feed off and give nothing back like the parasite it really is.

The GPL is a horrible choice for a free software license, the GNU project produces write only code and GCC is a bloated mess that is bested by clang in terms of memory usage, error messaging and speed. Coincidence? I think not. I don't notice a difference between clang and GCC in terms of the binaries it makes, all I know is that GCC hogs a lot of CPU time, memory and takes forever to build anything.

There's little wrong with proprietary software but I disagree that piracy of it actually harms the business. Open source software is great but I avoid it if it is released under the GNU GPL as that is a Marxist license. Marxism is a great theory but it's unrealistic and never worked in the real world. I prefer a truly free license like MIT/BSD/ISC. I have a goal of one day making a GPL free desktop. It will happen.

[Luke_Wolf]

TeamBlackFox, and FatBSDLoser Out!

unless you actually have something useful to add as opposed to Linux vs BSD and GNU vs BSD bashing we're trying to have a discussion here about the effects and benefits of permissive vs copy-left licensing, and we would rather appreciate if your crap didn't spill over here thanks. Go make your own thread where you are flaming each other if you must, but leave this one alone.

[bridgman]

So just to mess up this discussion and get all the warring parties onto the same side (against me) I'm going to propose that the ideal open source license would allow proprietary use for a certain period of time by a company/individual who contributed a certain amount to the project, with reasonable safeguards against abuse.

[Luke_Wolf]

The problem with this train of thought, and a lot of what Eric S. Raymond tried to claim over a decade ago, is that Open Source is not superior in every instance compared to other development methods. Let us not forget that while "free software" has been considered a muddled term due to its connotation in the English language with the foreign word "gratis", the term "Open Source" is also just as muddled and in some ways arguably more confusing due to it being applied both to software under a "free" or "open" license as well as being applied to projects which are developed under an "open" development model. While some projects that use an open development model do flourish, such as the Linux kernel, Apache, and other such projects, it has also proved itself to be terrible for certain others areas where a more tight control would be of benefit, such as game development. There are some examples were this does work, such as with Battle for Wesnoth which found a way to take thousands of different contributions and make it into a coherent whole, but in general it is far better to have a small closed development team for games, whether they in the end be released proprietary or under a free software license.

Another area from gaming which is actually in support of copyleft from the corporate perspective is the release of video game engines, such what John Carmack has done. Carmack choose the GNU GPL in order to keep development on these engines open but at the same time limit its effect on their engine licensing business. This way the community could do its magic to improve the companies older games while at the same time forcing people who wanted to use the code for their own proprietary game projects to still need to get a license from id Software.

The bottom line here is that neither copy-left or permissive is really any more "pragmatic" than the other, as the most practical choose is not always the same one in each and every instance.

Actually using gaming as an example against OSS is rather disingenous for one big reason. Nobody has ever tried to develop a serious game more complex than Wesnoth or Xonotic under an open source license. As a result there is no data on the matter as to whether OSS can succeed or not as a development model for games, and thus we can't actually draw any conclusions.

However! We do know that opensourcing engines has occurred to great success (For example look at the freespace engine and all of the high quality fan mods that were produced as a result the most high profile being Wing Commander - The Darkest Hour) , however all major instances so far have been with licenses that prevent commercial development.

Further! Situations where content is open sourced in all but license (For example Never Winter Nights, and The Elder Scrolls) have had great success, and the more open the content is to modification, the larger the modding community goes.

In Essence while the culture of gaming and modding is counter to open source licenses at this point in time the effects of more open systems and available systems on the associated communities is a strong indicator that open source development and licensing would be beneficial to the gaming ecosystem.

[zanny]

So just to mess up this discussion and get all the warring parties onto the same side (against me) I'm going to propose that the ideal open source license would allow proprietary use for a certain period of time by a company/individual who contributed a certain amount to the project, with reasonable safeguards against abuse.

Why is it important to give people the option to restrict the end users freedoms? What are they trying to hide? Obviously they are trying to do it to keep trade secrets, but doesn't that severely harm everyone else who has to either re-engineer their solutions or work around the black hole of binary blob they won't reveal?

I don't believe, at all, that the only way to profit off a novel idea is to keep it proprietary. Mediawiki is an open engine used globally, but wikipedia stands alone and gets tremendous funding by being the first major user of it, by being the primary sponsor of it, and by remaining open to everyone to use and contribute to.

Instead of saying "I have this idea, and a falsified economic model that enables me to restrict the distribution of information I produce, and also to restrict the necessary knowledge to know how said information operates, so I'm going to utilize this backwards system to try to profit off of discrete units by denying the purchasers the rights to know how it works and to modify it, I'm going to charge for what is expensive, ie, my time, and let people fund my work that way.

In a relevant example, if there were donation bins to get people working on radeonSI performance tuning, like "Full OpenGL 4 support will cost $50k to implement, divided amongst 5 developers over 2 months each", I'd throw $50 at that (more if I had it), and hopefully another 1000 people would (or 10k people giving $5) to make it happen because they want it and it would be valuable to the community to have. They would have to of course pitch it to me in a way that made me think it was feasible to do, and I'd like some kind of collateral if it doesn't work out, but it is a lot more organic and real than "here is a proprietary solution, give me money, and you don't get source, distribution, or modification rights because you live in a state that will point a gun at you if you try it without my permission". That is probably a bad example - better would be "10k to debug and fix why the 290/5/x don't work on radeonSI, 2k for parts, 4k a month for 2 months of labor". And of course it might not be funded - maybe someone else will offer to do it for cheaper. Maybe there just isn't any real demand for it anyway. But it would be open and real demand, not fake artificial scarcity.

I always worry internally if AMD can't gage the value of the gallium infrastructure because it isn't an immediate or direct revenue generator. We aren't really paying the AMD devs at all directly to work on it, and there isn't a survey I fill out saying what driver I use when I buy a GPU. At least Intel has the grandiose dream of having Linux tablets and steamboxes under their hardware domain, but AMD has catalyst competing for resources with the open team, where one has a really bright future and the other keeps trying to climb back up a cliff of shit that is infinitely tall.

But this whole nonsense of trade secrets and trying to extract artificial per unit value from something entirely imaginary and fake is ridiculous. A thousand years from now children will have their brains imprinted with the collective human history and they will find it hilarious that in the 21st century the ape folk were pointing guns and trying to hide their knowledge from each other for having the gall to try to share and collaborate on shared knowledge and scientific innovation.

As a result there is no data on the matter as to whether OSS can succeed or not as a development model for games, and thus we can't actually draw any conclusions.

0ad? It got kickstarted a fair amount, the developers are getting paid to work on it, and it shows in the amazing progress its made in the last year. I threw $40 at it, the price I'd pay for a boxed copy at a store, but society as a whole benefits from having access to the game code and art assets with open licenses. That is an obscene win/win for everyone involved. Can you imagine how much wasted effort there is remodeling and retexturing a barrel or a cardboard box in the hundreds of games that contain them because each company is a piss ant putting fences around their sandbox by not letting the other kids play in them over some misguided notion of IP scarcity. Once it is made it is immoral to hide it or restrict it, you just have to seek fair compensation as a scarce resource (like an artist, animator, or programmer) to do the making part.

[TeamBlackFox]

The main reason why I support the copy free movement and dislike the copy left movement is because I'd rather not stoop to the level of using copyright itself to prevent businesses from using my code. I see no issue with my code being used to make money, even if I am not credited. There is a saying "Mimicry is the highest form of flattery" so when people use my code it's showing they like something about it more, over that of a fellow developers code.

I just can't reconcile restrictions and the warped idea of freedom that the GPL promotes. Look at the BSD two clause license, and you'll see it has far more freedoms. I could only see the GPL work in the long term in some sort of fascist or non free market, otherwise a lot of code is wasted because if I ever want to use code that, lets say I added a commit to the Linux kernel to improve a certain system call, I would have to rewrite it entirely to commit it to FreeBSD or OpenBSD. And that's not fair, I own my code and should be able to do what I want with it. The other issue is that OpenBSD has a lot better code quality and I have seen a larger amount of interest from the set of developers I work with to contribute to OpenBSD because there are no strings attached unlike work towards the Linux kernel which, as far as I can see requires code to be GPLv2.

On the original subject of driving developers to UNIX and clones like Linux and such, I think that there needs to be support at the university level. Colleges back in the day usually used one type of OS throughout their infrastructure and professors would encourage their students to commit code to that OS. Now if you go to a university they have a load of Windows, OSX and Linux machines with no clear endorsement. In classrooms you need to use Windows for stuff like CAD, Visio and MATLAB, per the professors syllabus. If you have a BSD. OSX or Linux box and try doing the course they'll give you a zero if they can. Having professors who endorse UNIX and require their students to use it would go a long way to getting more developers out of universities to go and code for UNIX. The game classes focus on proprietary software like DirectX and not OpenGL and the like.