As much as the BSD developers should blame themselves, AT&T and Linus, I think most of BSD's lack of mindshare stems from the BSD license itself.
The BSD license encourages (or, at the very least, does not prohibit) companies from taking away stuff from the codebase, enhance it and run, never to look back until they need another software infusion. This does not make it easy for companies like Intel, IBM, Red Hat ot SGI to commit resources to the project, knowing that any other company can take what they developed and close it up. "Show me yours and I will show you mine".
GPL-like licenses prohibit such behaviour and are, perhaps, the reason why there is so much more GPL-ed software around than BSD-ed. That's why the same companies I mentioned before can feel comfortable that while others may use their Linux contributions, they will also be able to use everybody else's code forever. For me, using GPL for community-developed software is a no-brainer. It's not debatable it created a vibrant ecosystem around Linux.
The viability of BSD-ish licenses to create such ecosystems is, however, quite debatable. The Apache Foundation has several projects under such licenses that receive regular investment from big industry players.
Although I doubt there will ever be a clear-cut conclusion, it's fun to watch and participate on the debate.
More of the open source world does just fine without the GPL than you'd care to admit. The various BSD projects are also doing just fine without constant hype and Slashdot announcements.
> More of the open source world does just fine without the GPL than you'd care to admit.
60% of projects on Freshmeat use the GNU GPL. If you include other copyleft licenses, such as the LGPL, this is nearer 70%. -- http://freshmeat.net/stats/#license
Copylefted software includes such important projects as:
The Linux kernel.
Languages such as: C, C++, Objective C, Ada (all through gcc); Java.
Database engines: MySQL.
GUIs and toolkits such as: Qt, KDE, GTK+, GNOME.
GCC is particularly important, because nearly all open source projects are compiled with it or written in languages whose implementation uses GCC.
Look, I wasn't going to list out every popular open source project that doesn't use the GPL and I never claimed that there aren't useful projects that do use the GPL. I simply wanted to counter the mistaken notion that software without a GPL license sits in some small niche and all such projects have small inferior communities.
That being said, Freshmeat is a horrible way to support whatever point you were trying to make.
As I pointed out myself (and apparently you never got that far down my post) it is possible to form vibrant communities around BSD-licensed products. I even mentioned Apache as one of them.
My point was that BSD gets less investment because it's so easy to take away what you contribute and to use it to compete against you.
And thanks for the five pointers. Unless you can come up with a list two orders of magnitude larger, I consider my point fairly well demonstrated.
I don't think so. As much as free-software advocates like to pat themselves on the back, there are many projects out there that succeeded gaining popularity without a restrictive license like the GPL, and I also think that the GPL is so popular purely by fashion and because its restrictions are irrelevant for web-based applications.
A project's success can only be attributed to the people working on it, and to sheer luck.
Of course, it is kind of an incentive for companies to contribute back and for the community to have some legal protection, that's why I understand the need for LGPL or MPL. But GPL itself should not be considered open-source ... all the software you or I wrote is and will always be a derivate of public domain, and the usage of a library should not dictate the distribution rights of the final product, but only the distribution rights of the library in question. It's kind of ironic that proprietary platforms like dotNET are more open in this regard.
Also, I don't think Apple did more harm to open-source than Red Hat did. At least they don't pretend to be some kind of saviors of humanity.
> But GPL itself should not be considered open-source
The term "open source" has a clear, accepted, and well-understood definition -- http://opensource.org/docs/osd -- and the GPL clearly meets this definition. If you don't like the GPL, fine. But don't say that it isn't open source, because that's just not true.
If you choose to give your own private meanings to words, you risk being misunderstood. You also risk being seen as dishonest.
I know the term has rules for being used, but those rules are too permissive if GPL is accepted (not to mention GPL 3 which now includes hardware restrictions).
If I want to run away with a library and do my own thing, I don't see a problem with that if I'm giving back the changes to the library itself. Otherwise it cannot be viewed as a public good, as some might suggest. You can't compare a public good, such as a road to GPL software, because you aren't required to transport your merchandise for free on that road ... I know, stupid analogy, but you get my point.
Proprietary software shouldn't be considered a virus that we should get rid of, that's not an open mindset in my book.
The problem with Red Hat is that they sell a brand, not a product, and while many Red Hat employees are allowed to work on open-source stuff, there's no competition, no room for second place, no market. Unless you have a community of enthusiasts, like Debian of course, which is also technically superior in some ways, but those people don't see any direct earnings for their contribution.
Sorry, I have no clue what are you talking about. Can you elaborate? What /is/ the problem with RH?
You know that RH employees actually work on OSS, right? They're a major contributor to the open-source world, they have a community of enthusiasts (Fedora) and everything goes to upstream.
Look, I'm not questioning their contributions, just the current state of affairs.
If Red Hat vanished tomorrow, Linux would still be developed and maintained, even though it would suffer a setback. That's because there are many other companies and communities that help with that.
The problem is that only Red Hat earns serious money directly from selling Linux licenses, although it relies on contributions of many other companies and individuals to do so. The other companies only earn money from complementary products and consulting services ... services which don't scale unless you're a superstar developer charging $500 / hour or you have lots of resources to hire lots of people to answer support calls.
Although they shouldn't care since they are a company, and I do appreciate their contributions, the culprit is from my pov the GPL license ... and the perfect open-source license is LGPL-like ... that's all I tried to say.
The various flavors of Enterprise Linux are products. Since Red Hat provides paid support for Enterprise Linux, they have every incentive to make sure core Linux functionality is as stable and efficient to manage as possible.
You might say they don't want to streamline themselves out of a job, but there will always be demand for supported versions of Linux, no matter how easy it is to administer.
Another reason for RH to make sure Linux rocks is off course increasing the total mind share for Linux, for the more installations and applications there are, the better the prospectives.
intel, ibm, hp, and sun contribute to linux because it is the popular thing to do now, and it is in their interest to support it as an operating system because they all sell hardware, not (necessarily) because they like linux or use it. sun has its own competing, commercial operating system and yet they contribute to a rival one. why? because they can sell sun servers that run linux.
companies contribute patches, drivers, money, donations, and other resources to many of the free non-gpl projects because they use them internally and it is in their interest to keep those projects going.
bsd licensed code and operating systems have been around for decades. where are your examples of companies taking that code and "closing it up", making significant money off of that work, and without giving anything back?
> where are your examples of companies taking that code and "closing it up", making significant money off of that work, and without giving anything back?
Sun Microsystems and BSD? They have given stuff back, but basically, they did take BSD proprietary in a way that it took years to 'recover' from, in terms of a purely open source system.
This debate doesn't have one 'winner'. I suspect if you wanted to do some sort of rigorous, formal, analysis, you would turn to game theory. Barring that, though, I think that there are situations where the GPL is best: everyone can contribute knowing that no one will "run away with the pot", by, say, hiring up the core developers and taking the system proprietary. In other cases, BSD is best - with lots of software that gets used as a library, if it's under the GPL, even if companies want to contribute and be good open source citizens, they really can't, and so you lose any possibility of getting good work from companies. Even the FSF realized this and created the LGPL. So, like many things, it can't be boiled down to a simplistic "this one is better" type answer. It really does depend on the situation.
> * where are your examples of companies taking that code and "closing it up", making significant money off of that work, and without giving anything back?*
My understanding is that proprietary operating systems such as MS Windows use the BSD tcp/ip stack. Is this not true?
Many operating systems have networking API that is similar to BSD one, but that does not necessary imply that TCP/IP stack is same (and in case of Windows it is certainly not, as the BSD-like API is to some extent emulated in userspace).
On the other hand, some userspace parts of Windows are certainly traceable to BSD origins (i.e. some command line networking utilities and large part of POSIX subsystem userspace).
Windows had BSD code in their TCP stack for a couple releases of NT. There is still some BSD code in it and you can see the authorship notice if you do a strings on FTP.EXE;
I think Windows Vista got rid of the POSIX part by now. Embrace -> extend -> extinguish -> done.
Having BSD-derived userspace (like ftp.exe and probably more) does not imply having BSD derived TCP stack. And the internal architecture of windows networking is significantly different from any unix (one could probably also call it "broken") so I find it hard to believe that there are major parts of BSD-derived code in there.
Vista is probably first version of Windows where Microsoft actually advertises the POSIX part, so it's there and probably it's here to stay.
I cannot prove Windows had BSD code in its TCP stack as I never had access to its source code, but I never implied it must have been so _because_ ftp.exe has BSD code inside it.
There seem to be numerous references to BSD code inside Windows TCP stack from the late 90s. Maybe it is only utban legend, but maybe it's not.
I agree that the GPL is vital for getting contributions from commercial entities. In my company, management reacts very strongly to the possibility of giving another company an advantage and getting nothing in return. If I release code for a BSD-licensed project, my company might be subsidizing competitors, other companies that use our code but don't release theirs. The GPL provides a legal guarantee that our competitors have to play fair.
The legal guarantee is actually all we need. I'm not sure why that carries any weight with the bunch of ruthless mother...s I work for, but there you have it.
"A lawsuit between Unix copyright holder AT&T and University of California at Berkeley was also partly to blame."
As I recall this was a big hangup to adoption and/or commercial release. All the hardware vendors were using AT&T licenses and there were $500 licenses for PCs so the market for FreeBSD wasn't so compelling as that for Linux became.
"When you haven't been involved, you don't know that Bill Joy threw something in late one night, or Dennis Ritchie wrote something up as an expedient. "
This is a pretty good example of why software should be released early and often. They managed to destroy all their momentum with an 18 month long development cycle.
You have to wonder what their primary goal was. Perhaps they pursued their primary goal (a modern Unix with elegant internals) very intelligently and diligently and hoped that their secondary goals (users, developers, and influence) would materialize as a result. Or perhaps their primary goal was to win acceptance for their idea that Unix internals are crufty and deserve rethinking -- and perhaps they achieved that goal.
If winning users and developers was their primary goal and not just a secondary hope, then they got spanked and sent home, but I don't get that impression from the interview.
The BSD license encourages (or, at the very least, does not prohibit) companies from taking away stuff from the codebase, enhance it and run, never to look back until they need another software infusion. This does not make it easy for companies like Intel, IBM, Red Hat ot SGI to commit resources to the project, knowing that any other company can take what they developed and close it up. "Show me yours and I will show you mine".
GPL-like licenses prohibit such behaviour and are, perhaps, the reason why there is so much more GPL-ed software around than BSD-ed. That's why the same companies I mentioned before can feel comfortable that while others may use their Linux contributions, they will also be able to use everybody else's code forever. For me, using GPL for community-developed software is a no-brainer. It's not debatable it created a vibrant ecosystem around Linux.
The viability of BSD-ish licenses to create such ecosystems is, however, quite debatable. The Apache Foundation has several projects under such licenses that receive regular investment from big industry players.
Although I doubt there will ever be a clear-cut conclusion, it's fun to watch and participate on the debate.