GNU versus BSD

[phillipsjk]

Well, in a school setting, the instructors have access to your source code: it is what they mark.

GCC doesn't strip variables and function names unless you ask for a "stripped" binary.

[masher]

As I see it, the GPL and BSD licenses represent two extremes. I understand the rationale behind the GPL, because it guarantees that any modifications to GPL'd code are contributed back to the community. The BSD license is not viral in nature, like the GPL is, but can't guarantee that further improvements are not kept in proprietary repositories. Are there any licenses that strike a good middle ground? Is that even possible?

yes, this. Is there license that allows for the incorporation of open source into propritery programs, while still releasing the code to my part of it?

[Berengal]

The LGPL comes close. Proprietary programs can link to LGPL code (dynamically at runtime. The user has to be able to exchange the library for a different but compatible one). You're still not allowed to distribute a modified LGPL program/library without also making the source available, but at least you can contain the viral aspect to only the modifications you have to make to the library, not your entire program.

If that's your thing.

[Pesto]

Yeah, I'm familiar with the LGPL, but how would it work in practice if a stand-alone application were released under it?

[hotaru]

Yeah, I'm familiar with the LGPL, but how would it work in practice if a stand-alone application were released under it?

if you want to use some code from it in another app, you can make it into an LGPL-licensed library, and then use whatever license you want for the rest of your app.

[Pesto]

Hmm. That sounds very nearly perfect. I don't really see a downside to that.

[makc]

GPL people tend to be really silly about this though, in that if a library is released GPL, they want you to either dynamically link it, or if you statically link, you have to release the whole thing's source.

The FSF's position is that dynamic linking is enough to require your program to be GPL'd.

ok, so I make this proprietary closed-source program, a.exe, that loads open-source WTFPL b.dll. Now someone modifies b.dll and re-licenses it under GPL. Does this mean I should now GPL my a.exe? Silly, isn't it. I believe they could only demand that if I distribute a.exe with modified b.dll, but as long as I don't, I'm fine.

This could be seen in action with every 2nd PDF printer, that needs Ghostscript to work but instructs you to download it elsewhere.

[phillipsjk]

The purpose of that position is to close a perceived loop-hole. GPL advocates are worried somebody may take their work and "close" it by adding compelling functionality under a restrictive license; possibly tied to some kind of hardware patent. Or more likely, what Apple did with BSD to make OS X.

What they do not want is for you to "work around" the license by moving the GPL portion into a "library," then calling it from your proprietary front-end. While you may find this alarming that they seem to be trying to impose a license on calling programs; in reality, there are still enough loop-holes in the GPL that it is not a problem:

• Mere distribution of GPL software with other software does not impose the GPL on other software on the same medium.
• The GPL is not an ELUA: it only triggers for distribution or the creation of derivative works not exempted under copyright law.

This means that you, as the software publisher can still distribute b.dll under the WTFPL. The fact somebody "forked" the license does not affect you. Further, because the GPL is not an EULA, the GPL does not automatically take effect if the end-user substitutes the GPL version of b.dll for the WTFPL version. The GPL only triggers if the user tries to distribute the result.

This is how the BSDs are able to ship with GCC, even though most of the software is under a BSD license: BSD is considered a distribution or collection of software. The GPL does not "contaminate" the BSD distribution any more than needed.

[phlip]

ok, so I make this proprietary closed-source program, a.exe, that loads open-source WTFPL b.dll. Now someone modifies b.dll and re-licenses it under GPL. Does this mean I should now GPL my a.exe? Silly, isn't it. I believe they could only demand that if I distribute a.exe with modified b.dll, but as long as I don't, I'm fine.

As I understand it, you're right. You have a licence to distribute the original version of b.dll, but you don't have a licence to distribute the new b.dll as part of a non-GPL program. A user could swap in the new b.dll for the old one, but you couldn't write a.exe such that they were required to do so for it to work (say, calling functions that are in the new b.dll but not the old one... but still shipping it with the old one and giving "upgrade instructions" or something).

[stephentyrone]

Or more likely, what Apple did with BSD to make OS X.

The BSD layer of OS X has not been closed. Apple's bug fixes and improvements are on opensource.apple.com, under a free software license. In fact, quite a few projects that originated at Apple (i.e. are not descended from BSD) are available there as well.

[phillipsjk]

I never said that Apple closed BSD itself.

The objection is that Apple leveraged other peoples' work (without compensation) to make their proprietary product. The BSD people don't object to this: they want the best code for the job to be used. They are morally opposed to imposing onerous conditions on other programmers, so don't require "tit for tat" license terms.

However, the GPL people would ask:
Wait a second, why aren't the free versions of BSD the most popular ones?
How much lipstick do you have to add to a pig before people are willing to submit to an onerous EULA and fork over a lot of money in order to use it?
Yes, Apple did a good job pushing an integrated platform that "just works" for the most part. The problem is that the work needed to make a saleable product from a popular "free" distribution is not that much. Why should third parties be allowed to impose onerous terms on end-users just because they did boring integration and grunt work? GPL supporters believe the rights of users are more important that the rights of developers to choose any license they want.

Edit: Finally worked out how to include an RMS essay reference:

Non-copylefted software is vulnerable from all directions; it lets anyone make a nonfree version dominant, if he will invest sufficient resources to add significantly important features using proprietary code. Users who choose software based on technical characteristics, rather than on freedom, could easily be lured to the nonfree version for short-term convenience.

- The X Window System Trap

[Jplus]

Edit: Finally worked out how to include an RMS essay reference:

Non-copylefted software is vulnerable from all directions; it lets anyone make a nonfree version dominant, if he will invest sufficient resources to add significantly important features using proprietary code. Users who choose software based on technical characteristics, rather than on freedom, could easily be lured to the nonfree version for short-term convenience.

- The X Window System Trap

Circular reasoning. The arguments starts out assuming that the existence of non-free software is always a bad thing and then concludes that it is a threat.

I find copyleftism hopelessly moralistic. Let users decide for themselves!
I find Mac OS X worth the extra bucks because of the added convenience. I find iWork also worth the added bucks compared to OpenOffice.org, even though I still sometimes need OOo because iWork doesn't offer all features that OOo does (but iWork was not derived from OOo). I don't find Adobe Illustrator worth the extra bucks compared to Inkscape because I'm not a power user and because Inkscape works very well. Also I support Haiku and WebKit because I think those are good projects.
Users can recognise free-ness as a value in itself and take it into account in their choices along with price, features and convenience.

[phillipsjk]

I don't think you read the linked essay. It was prompted by a decision to change the License terms for X without consulting contributors:

This trust was misplaced. In its last year (~1998?), the X Consortium made a plan to restrict the forthcoming X11R6.4 release so that it would not be free software. They decided to start saying no, not only to proprietary software developers, but to our community as well.

There is an irony here. If you said yes when the X Consortium asked you not to use copyleft, you put the X Consortium in a position to license and restrict its version of your program, along with the code for the core of X.
. . .
In September 1998, several months after X11R6.4 was released with nonfree distribution terms, the Open Group reversed its decision and rereleased it under the same non-copyleft free software license that was used for X11R6.3. Thus, the Open Group therefore eventually did what was right, but that does not alter the general issue.

- The X Window System Trap

The point of copyleft is to force the developers and distributors to allow the users to decide for themselves (or was that sarcasm?).

[Jplus]

The point of copyleft is to force the developers and distributors to allow the users to decide for themselves (or was that sarcasm?).

With copyleft you force the developers and distributors to give the end users only one option, i.e. using open source software that in most cases is cheap but not as convenient and richly featured as it could be.

With copycenter you give the developers and distributors free choice. The consequence is that end users can opt to use either cheap open source variants of the software or convenient, richly featured variants, to taste.

[stephentyrone]

Yes, Apple did a good job pushing an integrated platform that "just works" for the most part. The problem is that the work needed to make a saleable product from a popular "free" distribution is not that much.

If the work is "not that much", then why hasn't anyone else done it? The reality is that it's an enormous amount of work. 25 years of development at NeXT and Apple by hundreds of full-time engineers. In the course of that work, besides the pretty user interface layer, Apple has made tens of thousands of bug fixes and performance improvements to the BSD layer and kernel, which are published as open source for other projects to use.

Why should third parties be allowed to impose onerous terms on end-users just because they did boring integration and grunt work? GPL supporters believe the rights of users are more important that the rights of developers to choose any license they want.

No onerous terms are being imposed on end users. The existence of OS X does not prevent you from booting FreeBSD. If you do so, you still benefit from the work that the engineers at Apple have done, because much of that work has been picked up upstream.

What you're claiming is that GPL supporters believe that they're are too good to get their precious little hands dirty with the hard work of actual engineering and that some shlub who isn't so clever as them and therefore doesn't have anything better to do should make anything "just work" for them but of course shouldn't be allowed to have any say in how their work is used because it couldn't possibly be a significant contribution. Most GPL supporters I know don't believe that, but that seems to be the line of argument that you're advancing.

[hotaru]

In the course of that work, besides the pretty user interface layer, Apple has made tens of thousands of bug fixes and performance improvements to the BSD layer and kernel, which are published as open source for other projects to use.

OS X doesn't use the BSD kernel at all, and the only changes they've made to the BSD userland are removing parts of it and a few minor changes to make things work with their kernel.

[stephentyrone]

OS X doesn't use the BSD kernel at all, and the only changes they've made to the BSD userland are removing parts of it and a few minor changes to make things work with their kernel.

Sorry, I should have been more clear. That's supposed to read (BSD layer) and kernel, not BSD (layer and kernel).

That said, if you believe that the only changes that Apple has made to the BSD userland are removing bits and minor changes, you're woefully mistaken. Many of the tools and libraries have essentially been completely rewritten over the years.

Let's pick one example, near and dear to my heart; the math library. The BSD math library from NeXT was unusably slow on PowerPC. Apple rewrote the core of it for 10.2, bringing in code from the old OS 9 sources (which were closed-source). Instead of keeping it closed, they released the much improved libraries as open source. Apple has continued to work on the library since then, to the point that OS X now contains almost no BSD-derived sources, is substantially faster and more accurate than both the BSD-licensed fdlibm and the glibc math library on both PowerPC and Intel, and is still published as open source.

[hotaru]

Let's pick one example, near and dear to my heart; the math library. The BSD math library from NeXT was unusably slow on PowerPC. Apple rewrote the core of it for 10.2, bringing in code from the old OS 9 sources (which were closed-source). Instead of keeping it closed, they released the much improved libraries as open source. Apple has continued to work on the library since then, to the point that OS X now contains almost no BSD-derived sources, is substantially faster and more accurate than both the BSD-licensed fdlibm and the glibc math library on both PowerPC and Intel, and is still published as open source.

http://opensource.apple.com/source/Libm ... el/s_cos.s and http://www.freebsd.org/cgi/cvsweb.cgi/s ... .S?rev=1.7 look almost identical to me... just like most of the other files in the math libraries on both operating systems.

[stephentyrone]

http://opensource.apple.com/source/Libm/Libm-315/Source/Intel/s_cos.s and http://www.freebsd.org/cgi/cvsweb.cgi/s ... .S?rev=1.7 look almost identical to me... just like most of the other files in the math libraries on both operating systems.

If you dig into the project structure, you'll find that the file you linked isn't used, except by the long-double cosl( ) function. The actual OS X / Intel implementation of double-precision cos( ) is here: http://opensource.apple.com/source/Libm ... incostan.c

Note that unlike the BSD implementation it uses an "infinite pi"* reduction, so you get accurate results even for huge inputs. It's also substantially faster on current Intel hardware. I assume when you say "just like most of the other files" you mean that you looked at the file names, decided they were similar, and moved on. The vast majority of the single- and double-precision implementations in the current OS X math library are original to Apple.

I'd prefer not to hijack this thread any further; if you'd like to continue the discussion, feel free to PM me and I'll be happy to discuss the provenance of the OS X math library at length.

*not really "infinite", just accurate enough that you get the same answer as you would if you had the exact value of pi. About 1200 binary digits. The reduction in the BSD library (using fprem1) uses 63 binary digits, by comparison.

[phillipsjk]

With copyleft you force the developers and distributors to give the end users only one option, i.e. using open source software that in most cases is cheap but not as convenient and richly featured as it could be.

Copyleft is a tongue-in-cheek term. By lowering the barriers to entry in the market, it is a capitalist's dream. One of the criticisms of Free and Open Source Software is too much choice. One of the FAQs many distros face is "Which which program is best for doing X?" Novice users have no good way to choose if they want (not a comprehensive list):

• Vim, Emacs, nano or Xedit for text editing
• Abiword, openoffice.org, or koffice for word proccessing
• oleo, gnumeric, or an office suite for spreadsheets
• lynx, links, iceweasel, or konqueror for web-browsing
• XFCE, KDE or Gnome for a desktop environment (I don't like those anyway: what happened to interchangeable window managers?)
• Plan 9, BSD,or Linux for a kernel

What a system integrator like Canonical or Apple can do is make reasonable choices for the novice users. If you a building hardware as well, you can choose either hardware that is supported or write any required drivers. Even if Apple used the GPLv3, they would be able to prohibit "hackintoshes" through Trademark law: no installing Apple branded OS on non-Apple-branded hardware.

If the work is "not that much", then why hasn't anyone else done it? The reality is that it's an enormous amount of work. 25 years of development at NeXT and Apple by hundreds of full-time engineers. In the course of that work, besides the pretty user interface layer, Apple has made tens of thousands of bug fixes and performance improvements to the BSD layer and kernel, which are published as open source for other projects to use.

I didn't realize that work on NeXT went back to 1985. Of course, BSD dates back to 1977, and the dump(8) utility dates back to "Version 6 AT&T UNIX," circa 1975. Of course over the years, those license terms have changed as well. I did not mean to imply Apple was doing anything improper using BSD code.

GPLv3

I like some of the newer, more restrictive clauses, but BSD people seem to hate it even more than the GPLv2... for the same reasons. I mentioned earlier that FreeBSD uses the GCC compiler. In reading the Freebsd-current mailing list, I noticed discussion about preparing to move to the clang compiler. FreeBSD core has decided to stay with GCC 4.2.1 because of distrust over the GPLv3. My Google-Fu has mostly failed to explain why GPL3 is such a problem. An e-mail message I came across suggested that many companies were wary of any GPL3 licensed software (ostensibly self-contained or not).

GPLv3 license marks GNU's decline was an interesting read, but I did not come away from that article understanding why the GPLv3 is worse than v2, other than increased legal complexity. The Wikipedia page of the GNU license lead me to Why you should use a BSD style license for your Open Source Project, which directly addresses some of my concerns:

A BSD license is not simply a gift. The question “why should we help our competitors or let them steal our work?” comes up often in relation to a BSD license. Under a BSD license, if one company came to dominate a product niche that others considered strategic, the other companies can, with minimal effort, form a mini-consortium aimed at reestablishing parity by contributing to a competitive BSD variant that increases market competition and fairness. This permits each company to believe that it will be able to profit from some while also contributing to economic flexibility and efficiency. The more rapidly and easily the cooperating members can do this, the more successful they will be. A BSD license is essentially a minimally complicated license that enables such behavior.

A key effect of the GPL, making a complete and competitive Open Source system widely available at cost of media, is a reasonable goal. A BSD style license, in conjunction with ad-hoc-consortiums of individuals, can achieve this goal without destroying the economic assumptions built around the deployment-end of the technology transfer pipeline.

I disagree with the last point: the Goal of the GPL is not to reduce the price of software to the "cost of media": the goal of the GPL is to promote freedom.

The GPL was designed to keep research results from transitioning to proprietary products. This step is often assumed to be the last step in the traditional technology transfer pipeline and it is usually difficult enough under the best of circumstances; the GPL was intended to make it impossible.

- GPL Advantages and Disadvantages section

This is true, however it is still possible to commercialize a technology without making it proprietary. Computer technology is a paradigm shift that I predict will take around 300 years to shake out. Current "State of the art" seems to be using DRM to make copying digital files "lossy" just like analog media. We are not even scratching the surface of what computers can do. Things like software patents and the wholesale conversion of "general purpose computers" to consoles stifle innovation.

I suspect much of the apprehension over the GPLv3 may have to do with this section:

3. Protecting Users' Legal Rights From Anti-Circumvention Law.

No covered work shall be deemed part of an effective technological measure under any applicable law fulfilling obligations under article 11 of the WIPO copyright treaty adopted on 20 December 1996, or similar laws prohibiting or restricting circumvention of such measures.

When you convey a covered work, you waive any legal power to forbid circumvention of technological measures to the extent such circumvention is effected by exercising rights under this License with respect to the covered work, and you disclaim any intention to limit operation or modification of the work as a means of enforcing, against the work's users, your or third parties' legal rights to forbid circumvention of technological measures.

Obligations concerning Technological Measures
Contracting Parties shall provide adequate legal protection and effective legal remedies against the circumvention of effective technological measures that are used by authors in connection with the exercise of their rights under this Treaty or the Berne Convention and that restrict acts, in respect of their works, which are not authorized by the authors concerned or permitted by law.

- WIPO Copyright Treaty (adopted in Geneva on December 20, 1996)

To me, that is a bit of a statement of fact rather than an onerous licensing condition. Since DRM can never be an "effective technological measure," it is impossible to violate that section of the treaty (barring Pi=3.00 type of legislation). However, I can understand why companies that have not yet given up on DRM may want to ban all GPLv3 licensed software from their premises. Note that the second half of Article 11 does not appear to rely on "effective technological measures"; authors are simply allowed to write their own copyright law. That is where Article 22 becomes relevant.

The section on patents (Section 11) seems to be saying you can't use patent law to restrict modifications to programs licensed under the GPLv3. IMO, section 7 of the GPLv2 is similar (except the license is revoked if a patent issue comes up). Edit: Okay, I admit I was wrong here: the "non-exclusive, worldwide, royalty-free patent license" is a legitimate concern for companies hoping to enforce patents in the future. I still think software patents are a bad idea.

The code within the src/external/gplv3 directories may have serious
legal impacts if you are a company and redistributing or changing
this code (as a company holding patents). We recommend you contact
your lawyer before using it.

Please do not import new GPLv3 projects without Board approval.

- http://cvsweb.netbsd.org/bsdweb.cgi/src/external/gpl3/?only_with_tag=MAIN (The README goes into more detail.)

Perhaps the most damning criticism of the GPL is the existance of the LGPL licences:

Which license is best for a given library is a matter of strategy, and it depends on the details of the situation. At present, most GNU libraries are covered by the Lesser GPL, and that means we are using only one of these two strategies, neglecting the other. So we are now seeking more libraries to release under the ordinary GPL.

Proprietary software developers have the advantage of money; free software developers need to make advantages for each other. Using the ordinary GPL for a library gives free software developers an advantage over proprietary developers: a library that they can use, while proprietary developers cannot use it.

Using the ordinary GPL is not advantageous for every library. There are reasons that can make it better to use the Lesser GPL in certain cases. The most common case is when a free library's features are readily available for proprietary software through other alternative libraries. In that case, the library cannot give free software any particular advantage, so it is better to use the Lesser GPL for that library.

- Why you shouldn't use the Lesser GPL for your next library

So, if I am reading that correctly, you only need the "freer" license if no "less free" alternatives exist? That is Bass Ackwards!

The comment about money is also telling. They seem to be implying that you can not make a lot of money releasing free software.

(Dang Forum timed out on me! Copy&paste saves the day!)

Edit2: removed spurious close-quotes added automatically when I closed the new quote with an open-quote tag.
Edit3: made this non-body text smaller.

[Jplus]

If the GPL is meant to promote freedom, then why does it restrict it more than necessary?

Your last quote from GNU.org suggested a "us against them" perception of open source developers versus commercial developers. There is no reason to present the state of affairs in that way, in my opinion.

[phillipsjk]

IMO, the GPLv3 is "restrictive" because the laws governing software are restrictive. You can get more jail time for circumventing DRM than you would get for killing someone. Section 3 explicitly says: ~"this software is not part of an effective DRM system, so changing the software does not constitute circumvention of an effective DRM system."

Are there any specific restrictive clauses you are concerned about?

No onerous terms are being imposed on end users.

Developers are free to take or leave GPL'd software. Many companies refuse to even give you their source code. I don't see a lot of developers complaining that Microsoft (as an example) is imposing restrictions of reverse engineering , or attaching restrictive licenses to many of their source code releases.

IMO, "copy-left" faces a double-standard because people equate "free" with the "public domain." As we know, people and companies are generally allowed to take things in the public domain and make them proprietary. On a related note: copyright now lasts longer than the medium the work is stored on (fifty years after the author's death in Canada). There is a very real possibility that many works from the turn of the 21st century will never enter the public domain (legally).

Edit: I can't believe I missed this: proprietary != commercial.
The linked definitions basically say that (proprietary == private) and (commercial == intended for the mass market). The FSF feels* that keeping the algorithms used by our computing devices secret is evil. Commercial software that is open can and does exist. The linked GNU.org text only mentions commercial software once: "University projects can easily be influenced; nowadays, as companies begin to consider making software free, even some commercial projects can be influenced in this way."

That does not seem to be an "us vs. them" attitude attitude towards commercial software vendors at all: the FSF wants them to release free (as in freedom) software. The "us vs them" approach only applies to "Proprietary software developers," which can, but does not necessarily include commercial software developers.


*Not an official FSF statement

[stephentyrone]

IMO, "copy-left" faces a double-standard because people equate "free" with the "public domain." As we know, people and companies are generally allowed to take things in the public domain and make them proprietary.

Neither people nor companies can take anything from the public domain and "make it proprietary". If I use something in the public domain, even if I use it for something proprietary, it is still available in the public domain for you or anyone else to use. My use of it does nothing to change that.

[phillipsjk]

• Companies are granted permits to harvest natural resources on crown land, or from International Waters.
• It is possible to buy a rare artifact (such as a Beethoven manuscript) at auction. For some reason, the act of cleaning up and reproducing the manuscript allows copyright renewal.
• Modern copyright law prohibits making copies of DRM-encumbered media before it degrades.
• Private vehicles are allowed to temporarily occupy public roads for the transport of goods and services.
• Utility companies are granted exclusive use of "rights of way" in exchange for relatively neutral or regulated customer treatment.

Computer technology is not isolated from the real world. Information can get lost or damaged. Of course, for popular projects, you are correct:

Only wimps use tape backup: real men just upload their important stuff on ftp, and let the rest of the world mirror it

- Found on Wikipedia

[styrofoam]

This is, of course, ignoring the fact that should the FSF ever be owned by or manipulated by another entity, you as a coder are fucked, and you can kiss the spirit of the GPL goodbye.

Correct me if I'm wrong, but isn't the FSF a legally recognized charity that is legally required to conform to their stated purpose?

[lalop]

I honestly don't see the issue behind there not being able to be a bsd -> gpl -> bsd flow. Anyone who BSD licenses explicitly gives everyone permission to relicense and not contribute back; if you don't like that, then.. use a copyleft license instead.

This is, of course, ignoring the fact that should the FSF ever be owned by or manipulated by another entity, you as a coder [who allowed relicensing under all future versions of the GPL] are fucked, and you can kiss the spirit of the GPL goodbye.

This is true (and I'd give props to anyone who has a youtube clip of Stallman answering this question) but, the way I see it, it's the author's risk to take. If you don't trust the FSF, then fine, just don't do it.