Software w/ Source for Sale?

frambooz asks: "As the GNU public license (amongst others) describes, you can make software that is free (as in freedom), but you don't have to make it *free* (as in free beer). I'm wondering if industry officials are aware of this fact, however. Do you know of any software packages that are Open Source, but still require you to purchase them? Did you ever work on such a project as a programmer yourself? If so, how did the development differ from a free(dom)/free(beer) Open Source application?"

Comments

Redhat EL 3?

[QuantumRiff]

You must buy the enterprise versions of Redhat, and you get the source, but not the source to some of their proprietary stuff.. ..

Re:Redhat EL 3?

[dtfinch]

You can rebuild a complete RHEL3 (minus the few bits of proprietary stuff I haven't noticed yet) from the public source, so long as you change the name and remove any Red Hat trademarks.

Some RHEL3 based distributions:
http://www.centos.org/
http://whiteboxlinux.org/
http://taolinux.org/

How about Qt?

[swillden]

Do you know of any software packages that are Open Source, but still require you to purchase them?

How about Trolltech's Qt for Windows? It's strictly commercial, but comes with source code. Actually lots of commercial development tools and libraries fit this model, probably because programmers find tools that provide source code to be much more useful. It's also quite common among certain software packages used by big business: packages that have to be heavily customized. Every large-scale point-of-sale software package (e.g. the software that runs the checkout lanes at the grocery store) comes with a source license, or makes one available for a reasonable fee (which everyone buys as a matter of course).

Binary-only software is a very new thing, historically. It was really quite uncommon prior to the rise of microcomputers, probably because the people who ran and managed the workstations, minis and mainframes were largely programmers and they found that software with source was more useful (who'da thunkit?).

If you're referring to software that is licensed under an open source license that allows redistribution, then no, I don't think you'll find too many packages that fit the model, and I don't think authors who try to do that will be immensely successful at selling software. But selling software with source works just fine, thanks to the power of copyright law. And selling services and support around truly Free Software works pretty well also.

Actually, I think that people who expect copyright protection for binary-only software are abusing copyright. The purpose of the law is to promote the growth of science and useful arts by encouraging publication, so that other people can learn from the ideas to create even more stuff. But these days we allow people to eat their cake and have it too; they can both obtain society's help in protecting their source code from illicit copying *and* they can also keep it secret to prevent people from learning their ideas. IMO, if you don't want to publish source code, you shouldn't get copyright protection for your source code. Use trade secret law to protect it. That's not as good, because under trade secret law if someone leaks it you can only go after the leaker, not anyone else who distributed the now-public information, but that's the tradeoff you should have to make: If I publish my ideas so others can build on them, then society will help me make sure no one copies my code, or creats unauthorized derivative works. If I keep my ideas secret, then I have to protect my code myself (I can still get copyright protection on the published binaries, however).

Unfortunately, old copyright law never foresaw that it might be possible to publish your work while simultaneously keeping it secret, and new copyright law has forgotten all about the need to balance private vs. public good.

Re:How about Qt?

[swillden]

I think that you may mean patents here.

Patents and copyrights both arise from this same underlying theory, and the same bit of the US Constitution. There are differences in the legal implementation because it makes sense to strike a different balance between the public and private interests for mechanisms than it does for expressions.

The reason copyright law has never explicitly required publication (unlike patents) is that without publication there was no way the author could commercially exploit his/her work, so there was no need to make that part of it explicit. The monopoly was intended to provide a measure of control to people who wanted to publish, since the nature of expressive works (writings, music, etc.) is such that once the material is published the creator loses all control of it, absent legal protections. Then software came along and it made sense to publish binaries and keep the source secret. One of the key assumptions made by the copyright laws -- but not the theory under the laws -- was invalidated by new technology.

In fact there *was* a sort of publication requirement to copyright law up until 1976: To gain protection you had to register your copyright and to register you had to file a copy of your work with the Copyright Office, who made available to everyone via the Library of Congress. That is actually still true except that as of 1976 you only have to register if you're going to litigate and you only have to file the first few pages and last few pages of any long work. The changes were primarily made to keep the Copyright Office from being buried in the ever-increasing flow of copyrighted material, but a side effect was that it made it possible to obtain copyright protection for something that is not published at all.

And, of course, something that is never published will never fall into the public domain, so why should the public protect it the author's monopoly on it? The author has kept it secret, so the author should just maintain control that way. Trade secret law even provides some enforcement tools.

Patents always explicitly required publication, because it was always possible to use/sell a mechanism without publishing it.

Re:How about Qt?

[swillden]

So you think Patents and copyrights are USA inventions then? Bloody arrogant yanks. From the constitution !!??? arrrgh!

Of course not. But I was talking about US law. Sorry for the America-centrism, but slashdot *is* an unabashedly American web site, even if there are visitors from all over.

But, to give credit where credit is due, copyrights were actually invented by the British primarily as a means of censorship. This type of copyright started out as sort of an agreement about publisher's rights (publishers owned what they printed, not authors) and was formalized in the Licensing Act in the mid 1600's (sorry, don't remember the year off the top of my head) which gave a monopoly on all printing to the Stationers' Company of London. In exchange for this monopoly, the Company agreed not to publish anything the Crown didn't like.

The first modern copyright law, which attempts to promote authorship, also originated from the UK, in the form of the Statue of Anne, passed in 1709. It gave copyrights to authors and established the term of 28 years, which the young United States adopted nearly 100 years later.

The origin of the the basic structure of the copyright law in effect in the US and around the world today was defined in the Berne Convention in the late 1800s. The Berne Convention established copyrights that are automatic, without the need for registration, and last the life of the author plus 50 years. The US didn't fully adopt this approach to copyright law until 1976 and 1998.

All of that said, in the US, the power of Congress to establish copyright laws arises from a specific passage of the US Constitution. Further, the philosophy underlying US copyright law was, for nearly two hundred years, really defined by a series of letters between Benjamin Franklin and Thomas Jefferson, who defined the basic social contract that US copyright law has been based upon. This American view of copyright had subtle but important differences from the British and European views, based on the relative importance early Americans placed on freedom of speech and action. The American view of copyright was much more skeptical of the value of granting monopolies. Was. No more, unfortunately.

Now, should I close with some snide comment about knee-jerking Limeys? Nahhh.

Dont confuse GPL and open source

[prostoalex]

Well, that's how the things usually work in the embedded world from the startof the industry. In the embedded programming you do need to customize the purchased packages occasionally, as you switch to different architectures, different chips within the same architectures and so on.

So you buy a software product (TCP/IP stack, or some drivers), and they usually come with the royalty-free source. You cannot resell the product, but you can embed the product in your own system and resell that. The source belongs to the original manufacturer and is free for you to tinker with.

That's kinda the way the things always were in the embedded world, too many people are confusing the GPL and GNU ideology with the concept of having sources on additional CD with the product you purchased.

Re:Dont confuse GPL and open source

[silicon not in the v]

Yeah, I think the poster of the original article is not clear on the term "open source" that he uses. He indicates this in wondering if "industry officials" are aware that they can charge for open source stuff per the GPL. He is probably thinking of what I would call "source included". That would not be under an open redistribution license, as the GPL is. Most companies would not go for that because there's no way to make people buy it anymore.

selling a product vs. download

[urdine]

The Internet lays bare an interesting human trait - that people are willing to pay for a product in a physical form that they could grasp from the ether for free.

Examples: People who bought RedHat in the store for the box, the manual, and the CD, rather than downloading it. Same thing when people buy the PDF version of an online article so they can print it out or just "keep" it - people do this more than you would think.

I think there is real value added by having a physical product, and some of the bigger open source software projects should do this if possible. Not only is it another way to get funding for the project, but it's actually something a lot of people want. Wouldn't you like to have Wikipedia 1.0 sitting over your desk?

GNU Emacs

[gorre]

The earliest example of selling free software is probably RMS selling tapes of GNU Emacs for $150 a tape. He says he sold 8-10 tapes a month which generated enough cash for him to live off.

Re:Forgive my ignorance, but how?

[(H)elix1]

How can you give away the source for free, yet also charge for the software? What stops people just downloading the source and compiling it, without paying?

The build process for some things is not trivial... Some of the big distros - SuSE and Redhad enterprise versions don't (and I could be wrong, but bare with me) have the source code available. Reproducing those distros in the binary bootable iso format is not for the faint of heart. Look at the elbow grease it took to get White Box Linux - compiled from the source of RHEL - up and running. In corporate, it is often easier to buy open source kit than get it running yourself. As a bonus, you get someone else to take care of the maintenance....

Re:Some..

[QuantumG]

Have any games companies released a game that used the GPL version of the Quake 2 engine? It would seem to me that it's a pretty good deal: just stick a tar of the source on the CD and you've forefilled the requirements of the GPL. Of course, your content need not be distributed under the GPL.

Re:Some..

[drsmithy]

Game companies seem to understand it.

Game companies have the luxury of a significant part of their product being covered by the 'non free' license of copyright - textures, level maps, music, etc. Unlike most software, a great deal of the value in a game is *not* in the source code.

It's substantially harder to sell "free as in freedom" (how I hate that term) software when the only real component of value in it *is* the source code, because your first customer can then turn around and give away (or even resell) your product.

As an aside, I expect to see more companies starting to use a tactic like this to create products using "free as in freedom" software without having to give away a fully functional version in source form by integrating non-source-code "stuff" that isn't covered by the GPL. Theo's copyright on the OpenBSD CD structure is a primitive example. Copyrighted filenames (if possible ?) would probably also be a similar way of stopping someone else just grabbing your source code, recompiling it and on-selling your product without having to expend time themselves renaming all the source files.

Re:Some..

[Cryogenes]

Surely, a game using the Quake 2 engine is a derived work of that engine. So if you want to release such a game you either do it under the GPL or you pay id for a commercial license of the engine.

BTW, that is the same trick that Borland used for Kylix. You can use the free version of Kylix, but if you do, you can only distribute your programs under the GPL. If you pay for Kylix, you can distribute your programs any way you like.

Re:Forgive my ignorance, but how?

[(H)elix1]

One of the hazards of posting from a mobile phone... the 'guess to what I am trying to spell' does not always get it right (grin)

Re:Forgive my ignorance, but how?

[drsmithy]

The build process for some things is not trivial... Some of the big distros - SuSE and Redhad enterprise versions don't (and I could be wrong, but bare with me) have the source code available. Reproducing those distros in the binary bootable iso format is not for the faint of heart. Look at the elbow grease it took to get White Box Linux - compiled from the source of RHEL - up and running. In corporate, it is often easier to buy open source kit than get it running yourself. As a bonus, you get someone else to take care of the maintenance....

True enough, but keep in mind that "hard work" only has to be done _once_ by someone prepared to give it away and the value of that software disappears.

Dansguardian is a violation

[petrus4]

The author of Dansguardian states first that his software is not free for commercial purposes, but then says it is licensed under the GPL. This is a contradiction.

The GPL does not explicitly stipulate commercial or non-commercial use. Its only requirement is that source is distributed with binaries, and that all subsequent derivative works are also licensed under the GPL. If this guy wants to make his software non-free for commercial use, he should write his own license.

rogue wave

[rmull]

Rogue Wave Software always used to ship source. I assume they still do.

Bad Assumptions

[Brandybuck]

You're making erroneous assumptions, but don't feel bad, because GNU makes the same ones. The first bad assumption is that the reality follows theory. The second is that you're selling software.

The theory is that you can put a price tag on Free Software. The reality is that you're going to find precious few customers for it. Free Software is like a gate without a fence. You can charge all you want to use the gate, but without a fence you won't find too many people paying you for that privilege. You're going to sell to the first customer, and maybe to the second, but by the time you get to the fifth or sixth you'll find your earlier customers have become your competitors and driven down the market price to zero dollars.

There are no publicly available Free Software packages with a market price greater than zero. Take the GCC toolkit for example. GNU theoretically "sells" it for $45, but in reality everyone gets it for free. I know that I have several copies laying around, and I didn't pay for a one of them.

Which leads to the second bad assumption. When you find people "buying" Free Software, they aren't really buying the software. When you buy a boxed set of SuSE or Redhat you are not buying the software, you are buying the box, manuals, service and support, and the convenience of not having to download and burn your own CDs and DVDs. The software itself is free (as in french fries). Or take the "Deluxe GNU Distribution" which GNU sells for $5000. Do you really think people are buying GNU software for that kind of money? Of course not! They're buying a combination of custom prebuilt binaries and the warm fuzzy feeling a generous charitable donation gives.

Some people will be fooled, however. Some people will buy SuSE and Redhat not knowing that they can get it for no cost. What they're buying in this instance is an education. One notable commercial Open Source figure once admitted to me that he was in the business of taxing ignorance.

Please stop spreading this myth that you can sell the software. You can't. You can certainly sell a convenience, a support contract, or even warm fuzzies, but you cannot sell practically sell the software. The realities of economics won't let you.

Re:Bad Assumptions

[pete-classic]

You're making erroneous assumptions, but don't feel bad, because GNU makes the same ones. The first bad assumption is that the reality follows theory. The second is that you're selling software.

You are flatly wrong. Please read what the FSF actually says on the topic before you comment on what they think again.

Here are some things that the FSF actually says on the topic (emphasis in italics is mine):

Actually we encourage people who redistribute free software to charge as much as they wish or can.

With free software, users don't have to pay the distribution fee in order to use the software.

You can charge nothing, a penny, a dollar, or a billion dollars. It's up to you, and the marketplace, so don't complain to us if nobody wants to pay a billion dollars for a copy.

I don't think that the FSF suffers under any illusions about being able to sell Free Software in the way that commercial software is sold.

However, when people think of ``selling software'', they usually imagine doing it the way most companies do it: making the software proprietary rather than free.

So unless you're going to draw distinctions carefully, the way this article does, we suggest it is better to avoid using the term ``selling software'' and choose some other wording instead. For example, you could say ``distributing free software for a fee''--that is unambiguous.

They seem to make some of the the very same points you are trying to make. So please, chill out, and don't speak for other people until you are sure what their position is.

-Peter

PS: GNU is a system, the FSF is a group. GNU doesn't make assumptions, the FSF does.

Re:Forgive my ignorance, but how?

[Yaztromo]

How can you give away the source for free, yet also charge for the software? What stops people just downloading the source and compiling it, without paying? Or do you just rely on people/companies being good enough to CHOOSE to pay for it, when then don't actually have to?

If you're distributing an end-user product targeted towards non-techies, many of them simply don't want to be bothered with having to compile and set-up everything. They don't want to have to ensure they have a compiler installed (remember: other than Linux/*BSD/OSX, most consumer OS's of the last 10 years don't come with a compiler), and all of the necessary pre-reqs to do the build. They want something that comes packaged inside a nice installer that just works, and in many cases couldn't care less if they could get it for free if "free" means having to go through a whole pile of complex mumbo-jumbo they don't care about.

That's the situation my project, the jSyncManager is in. To build everything, you need the Java2 SDK v1.3 or better, Apache Ant, the Java Communications API, the Java USB API, the jUSB API, jDOM, Java Help, Java Mail, and Java Activation. To run the jSyncManager, you need many of the same APIs installed (although we've coded things to make some of these packages optional -- if you don't have Java Help installed, for example, the Help option is simply greyed out). You can download the three JARs for the jSyncManager, and put them into the right directories and create the necessary desktop icons yourself if you want -- but many of our potential users get confused at "Java" -- nevermind all the rest.

We do have some bundles with all the right cross-platform APIs present and accounted for, and are hoping to put together some platform-specific bundles with nice installers and the platform-specific libraries (like javax.comm and javax.usb) -- but it's a lot of work. We have lots of users (and potential users) who would be happy to pay a small fee to have us put everything together in an easy-to-install package so they don't need to know the underlying magic incantations, and they don't have to go around to half a dozen (or more) different websites to download and set-up pre-reqs.

Sure, advanced users will simply download the parts and put things together themselves, paying nothing -- that's what I expect them to do. But there are people out there willing to pay for convienence as well. And why shouldn't I make a bit of money to give them what they want/need?

Brad BARCLAY
Lead Developer & Project Administrator,
The jSyncManager Project.

Re:nononono

[swillden]

The Kylix trick is inherently flawed and probably contradicts the GPL IMHO.

No, there's nothing wrong with it. You can't really build a Kylix app without using Borland libraries, so your app will include Borland code, making your code a derived work of theirs. If you don't have a license to distribute their code you can't distribute your app. Since they release their libraries under the GPL, you can release your app under the GPL, if you like. If you want to release under some other license, you have to pay Borland for a different license for their stuff.

Other software packages like Qt for X11, Sleepycat's dbm implementation, etc., work the same way. It's a pretty nice model for development tools, although it wouldn't work for most other kinds of software.