We Don't Need the GPL Anymore 919
jpkunst writes "In a lengthy interview with Eric S. Raymond by Federico Biancuzzi at O'Reilly's onlamp.com, ESR defends his position that 'Open source would be succeeding faster if the GPL didn't make lots of people nervous about adopting it.'" From the article: "I don't think the GPL is the principal reason for Linux's success. Rather, I believe it's because in 1991 Linus was the first person to find the right social architecture for distributed software development. It wasn't possible much before then because it required cheap internet; and after Linux, most people who might otherwise have founded OS projects found that the minimum-energy route to what they wanted was to improve Linux. The GPL helped, but I think mainly as a sort of social signal rather than as a legal document with teeth."
Re:He's right, of course (Score:5, Informative)
While shipping NT 3.1 Microsoft was under pressure to add TCP/IP so they bought a commerically available stack rather than write it themselves. This commercial offering was a BSD derivative -- completely legally. For NT4 Microsoft rewrote the stack substantially, retaining old bits for backward compatibility. If this is 'stealing' from BSD, we should just scrap the BSD licence since it's not worth the paper it's printed on.
If this were true... (Score:3, Informative)
I think it's clear that the reason most open-source developers are inspired to work on Linux is the knowledge that their work won't be commercially exploited.
Re:Just how much of the document has teeth, anyway (Score:5, Informative)
Shortly, it works like this: Company Foo infringes the GPL. If they go to court, they can try to argue the GPL doesn't apply - bad idea, since now it's entirely a copyright matter. And copyright says you can't take somebody else's stuff without permission, which means they're screwed.
Here's the thing, the GPL is the only thing that gives you the permission to redistribute the code. If you don't like it, that's fine, nobody forces you to agree to use it, but then the whole thing falls back to copyright law, which doesn't give you the permission to redistribute anything.
The GPL is unique in that it *grants* you privileges, instead of taking them away. Fighting the GPL will result in losing those privileges.
That's why nobody goes to court, because they wouldn't even be talking about the GPL there. They'd be deciding if there was or not copyright infringement.
Re:Other licenses are becoming more common (Score:1, Informative)
Re:GPL Teeth? (Score:3, Informative)
Re:GPL Teeth? (Score:3, Informative)
No, more often than not, the GPL software is a small component of a larger system (i.e. code to handle graphics formats). The GPL'ed code is not changed or improved, but somehow the communitiy expects to get all the other irrelevant code for free. This is what turns companies off to it.
You don't have to open everything. Just the stuff that is a derrivative of the GPLed program.
You do if it is statically linked. Of course you can decouple the GPL code from your application, but you might take a penalty somewhere (in size, speed, or complexity). Most companies are not going to hassle with that and just pay someone $30,000 to write the widget.
Re:Mac too (Score:2, Informative)
Wrong. Close... but still wrong.
OS X uses a Mach microkernel with a BSD compatability layer.
Which, as far as most users are concerned, is pretty much the same as saying "it's BSD," but under the hood that's not exactly true.
Re:Other licenses are becoming more common (Score:3, Informative)
No sir. Perl [perl.org] and MySQL [mysql.com] are GPL'd.
Re:GPL Teeth? (Score:5, Informative)
You don't have to open everything. Just the stuff that is a derrivative of the GPLed program.
And even that's not entirely accurate. If you take GPL'd code, modify it, and use it in house, you don't have to release it. The ONLY time that you'd have to release code is when you're distributing a derivative work. For example, if you modify code, and then turn around and sell it, when you sell it, you also have to provide a copy of the source to the people who buy it. If you release it for download, you have to release your changes. That's it.
~Wx
Re:Amazing (Score:2, Informative)
In the real world, where 90% of commercial programming is done in-house to create in-house applications, no license comes closer to meeting corporate requirements
I used to work for a large multinational retailer that did most of its development in house as you mention for in house applications.
We were under restrictions from legal when using GPL'd code because there was no clear definition of 'distribution'. According to some of the legal review of the GPL, we would have been distributing our code when we made it available to solely owned affiliates of the parent company. They were part of a separate business line with their own IT and executive staff, more a maintenance organization than a retailer (they did not have stores).
As an overall enterprise we were consolidating on common platforms (bulk purchasing power) across the board, but were still separate entities. The legal advisors indicated that our sharing of code with GPL components to these affiliates consituted distribution which would have activated the viral nature.
Whether this is accurate or not IANAL and can not answer. All I know is this very large organization with a large legal staff determined it was a possibility, and therefor restricted us to not use GPL code unless it could be alternately licensed or was supplied as part of a software purchase from a vendor that had the liability.
Re:It's not the GPL that makes people nervous (Score:1, Informative)
Wrong. (Score:2, Informative)
If he honestly doesn't believe that the GPL has helped get Linux and all of the tools it needs to where it's at, he's a fool. We're still regularly finding people that aren't following it.
The reality is that we're entering the post-secret code world. The license doesn't matter so much, people want source code from Sun, IBM, even Microsoft because it makes their investment more secure. It's only the fringes that really care about the licenses that much. There are some opponenets to the movement who will bitch about the GPL (Sun, MS..) but it's simply reaching a point where that doesn't matter too much, nobody who is honest about business really cares to steal someone else's code. Nobody who is really serious thinks that they can get away with it. It's really about being able to maintain your investment and possibly customize.
Eric, you should spend your energy debunking the GPL detractors rather than spreading their FUD. It's really pretty simple, if you want to keep your code secret, then write it yourself and do that. If you want to play with others then be willing to share with others. If you're making a project that is primarily GPLed code, then maybe you should think about it before you try to call it your own stuff and keep the code secret, you really don't have much of a competitive edge in the first place.
Re:-1 Flamebait (Score:2, Informative)
ESR once described himself as "one of the senior technical cadre that makes the Internet work, and a core Linux and open-source developer", which is so mind-blastingly far from the truth that I've taken nothing he's said seriously since.
Re:He's right, of course (Score:1, Informative)
Re:GPL is very much needed (Score:3, Informative)
Re:A consultant's perspective (Score:2, Informative)
I want to first emphatically state (for those who are clueless) that whoever told this company that they had to GPL their entire project was a fuckwit.
You are allowed to use a GPL'd Java driver in your proprietary software as long as:
And now I wander into the tangent of "what the hell were the programmers doing?". The idea I try to live by is that I don't give a flying fucking shit what the database and drivers are. I create a table structure and an automated way to create those tables in a SQL database; I create code that goes against a JNDI data source; and I test against SQL Server, Oracle and MySQL. Maybe DB2. Unless it is required that you do a lot of things SQL side (such as triggers on the database side), the idea is you abstract out the database dependencies and let the customer choose what database they want to support (and so what drivers they need to use). But that's just another consultant's perspective.
Re:Ugh... no (Score:4, Informative)
What parts are those, precisely? I don't know of any parts of Safari which remain proprietary that are any kind of barrier to competition, or that are technically difficult to implement. Safari is a very thin shell around Webkit, and there are at least two open-source replacement shells (Sunrise Browser and Shiira).
use of the LGPL in this case has still created an effective barrier to the open source product being as useful or successful as the commercial project which is using its code.
I'm completely unable to understand how you would come to this conclusion. Safari itself only uses standard Mac OS X APIs, so Apple could have open-sourced all of Safari (and Dashboard, but that came later) without open-sourcing any other part of OS X, no matter what open-source license KHTML or Webkit was released under.
About all that placing KHTML (and thus Webkit itself) under the GPL instead of the LGPL might have done would be to keep Apple from using Webkit in Mail in Tiger, and make some third party products on OS X use one of the other HTML rendering packages instead. The only program I can think of that I use, that uses Webkit, is Adium. And that's already GPLed.
So, Apple has in fact released all the code that is needed for a third party (be that the KHTML team or Nokia) to duplicate "the commercial project which is using its code", just as they would as if KHTML had been released under the GPL. Apple could have created the kind of barrier that you're talking about, but they chose not to.
Unless there's some magic Safari goodness that programs like Shiira are missing (and I doubt that, Shiira already does more than Safari) I'm completely at a loss to understand what you're getting at here.
Re:GPL Teeth? (Score:4, Informative)
The company I work for origionally had a CTO that was freaking out about GPL'd software.
It came to pass that he was simply acting that way because of the FUD and scaremongering that the MS rep he was buddy-buddy with was feeding him on a regular basis. After his "demands" and we presented a proposal for rewriting and purchasing everything needed to eliminate all GPL software in the business plus a letter from the Company's law firm telling him that the GPL is 100% harmless in every aspect unless we are shipping GPL code as or in a product.
The whining was still there, finance refused to approve a 2.2 million dollar budget line to buy all new MS and other commercial software as well as hiring programmers to rewrite from scratch some of the other solutions we rely on for revinue.
the GPL is not "dangerous" or "viral" and also is not scary in any way, shape, or form to anyone but someone trying to steal code, get something for nothing or are underinformed or relying on lies/bad information.
We proved it to a CTO that was pigheaded, must have his way, and trusts his personal friends more than the experts he hires.
Re:A consultant's perspective (Score:2, Informative)
I Call BS. The official MySQL JDBC drivers are GPL, but you can purchase them under a different licence for commercial use. Plus, there are 3 LGPL licenced JDBC drivers for MySQL that I know of, that you would not have to pay for OR release your code as GPL to use.
There is no way a company looked at the cost of licensing the MySQL JDBC drivers, and then decided on cost alone to go with Oracle! Oracle costs lots of money (and is arguably worth it) and MySQL licensing is peanuts next to that cost.
Re:BSD is a great example of what doesn't work (Score:5, Informative)
Yahoo!, Apple, and Pair Networks (in money) would probably argue against that.
The companies you mentioned probably use little if anything of the BSD code any longer.
I don't know of any major corporation which has made significant donations back to the BSD core. There may be the rare exception, but the bulk of corporate back-donations has been some bug fixes. That has left the development almost entirely to individual developers or very small groups, and thereby limited how much could be done.
Most companies hire contractors to contribute to the BSD's. You do not see many companies make big shows about it.
It has been my observation that the BSD source base has been relatively stagnant over more than a decade. If you look at what a modern BSD provides and compare it to what BSD 4.3 provided you'll find little that is new. A similar comparison with any major commercial UNIX will yield a great many such features (like working SMP support, journalled filesystems, NUMA support, logical volume management, realtime support, etc).
Remember the list of features modern UNIXen have that BSD doesn't? Did you notice how many of them Linux does support?
When will Linux support Soft Updates? When will Linux use sysctl() instead of
To be sure, one of the major limitations in the BSD codebase has been the reluctance of the BSD principals to accept code they didn't write.
Huh? That does not match to what happens within the BSD communities. Maybe, you are thinking about the Linux community?
Re:BSD is a great example of what doesn't work (Score:3, Informative)
Doesn't it already? [linux.com.hk] "A sysctl call has been present in Linux since version 1.3.57." However, Linux also lets you use the
How about virtual channels on a sound card?
ALSA has a method for doing that (the dmix plugin). However, it's not enabled by default, because for cards that actually have multiple channels, it's better to use the real ones.
Maybe, you are thinking about the Linux community?
No, I don't think he was. Seems like you both are mildly ignorant of the other camp (or were you purposely asking about thing that are actually in Linux? In that case, I apologize for being thick).
Re:BSD is a great example of what doesn't work (Score:3, Informative)
Working, but not great. And the threading support is weak.
Journaling is currently being added.
Which makes it irrelevant right now. Until it's been released and proven to be stable and reliable, I don't see any widespread usage of it. (Plus, with FreeBSD 5's less the spectacular record of working well, I wouldn't consider a new FreeBSD filesystem for quite some time.)
USB support existed in the BSD's--I believe NetBSD had it first--about two years before Linux.
This feature isn't too useful outside of desktop/workstation usage, so most corporations won't be too terribly interested. Furthermore, USB support does exist in both operating systems now.
Jails have existed in FreeBSD for quite some time.
This is something that is missing in Linux. However, its usefulness is limited; When will Linux support Soft Updates?
Linux as a whole never will. Soft updates are a feature of individual file systems, of which Linux supports many. Soft updates and journaling file systems are mutually exclusive, and given the success of journaling file systems, I don't see them coming to Linux any time soon.
When will Linux use sysctl() instead of /proc?
Many years ago? [maconlinux.net]
But even better: when will this make a huge difference in how either operating system works or in functionality?
How about virtual channels on a sound card?
Only a major concern for desktop users. Just about every sound card nowadays has hardware mixing; why not use that instead of your kernel-level (still software, not hardware!) mixer?
Many of the features you have listed are of no interest to many corporations, unless they are in the business of proving desktop operating system solutions.