A Blood Curdling Scream!

Just realized... This is probably a bad day to be confronting the establishment.  For those that understand, just smile and wish me luck.

Jack

Jack, Although they have started offering licences for MYSQL, that is mostly optional. Nobody is forced to buy that licence. And there are very little cases where anybody would want to or need to  to buy that licence.

MySQL :: Commercial License for OEMs, ISVs and VARs

Although they worded it a bit cryptic trying to try to include a few more cases, you would only have to buy the licence if you took the open source MySQL sources, added a few extensions for your own use-case, and then are selling the combination, while you're unwilling to provide sources for the extension that you made.

This case now requires a licence, which previously was either impossible or otherwise not publicly offered.

When e.g. you'd build a Linux-based "appliance" which includes a MySQL database, and you inlcude a few closed-source applications, then you still are NOT required to purchase a licence. This is the case that they are obfuscating a bit. (of course they didn't mean to).

I am afraid this is going to launch into a rather heavy discussion.  You pointed directly to the issue at hand. The line:

"This case now requires a license, which previously was either impossible or otherwise not publicly offered."

should say instead ...was not necessary.  The usage agreement of GPL (oh, by the way the "L" stands for license) is quite clear, which is a free use, (free like beer, not freedom), and if improvements are made, they are to be returned to the community.  That is the contract, and yes, it is a contract. If you do a quick search to get the history lesson, you will see that back when the GPL was being crafted, it was litigated and refined by a substantial number of lawyers at the corporate and private levels to make sure it was air tight.  To say that it requires a license is correct, but it has a license. To say that it was not offered is incorrect, because it was already in place (though not in the best interest of the company).

Jack.

What I mean with "this case" is: if you want to "improve" as you call it and THEN distribute your work along with the original GPL stuff, without providing source to your improvement. That is what the GPL disallows. As you say, the licence (GPL) allows a lot, but not that.

If the person or entity holding the copyright does not offer dual licencing and you still want to do that, it is not allowed.

The GPL says nothing about money. (Free as in beer). I can (and have) improve on GPL code, and sell it. GPL then requires me to provide the sources (if they ask) to anybody who got the binaries from me. GPL says nothing about putting it on the internet, or requiring my sale of the improved version to be "free". However, my first client is then (again according to the GPL) allowed to sell copies of MY improved code, or put it on the internet. This would kind of spoil my market. The GPL thus causes most GPL-ed programs to be "free-as-in-beer", because if you'd ask for even a small amount, the guy next door (and/or your first client) can do it for free.

Also note that the GPL is perfectly fine with me improving a GPL-ed program in-house and then "not giving back". But as soon as I give or sell the improved version to someone else, I'm required to provide the sources to my changes to them. And they get the right to redistribute / resell / publish the whole thing.

What I mean with "this case" is: if you want to "improve" as you call it and THEN distribute your work along with the original GPL stuff, without providing source to your improvement. That is what the GPL disallows. As you say, the licence (GPL) allows a lot, but not that.

If the person or entity holding the copyright does not offer dual licencing and you still want to do that, it is not allowed.

The GPL says nothing about money. (Free as in beer). I can (and have) improve on GPL code, and sell it. GPL then requires me to provide the sources (if they ask) to anybody who got the binaries from me. GPL says nothing about putting it on the internet, or requiring my sale of the improved version to be "free". However, my first client is then (again according to the GPL) allowed to sell copies of MY improved code, or put it on the internet. This would kind of spoil my market. The GPL thus causes most GPL-ed programs to be "free-as-in-beer", because if you'd ask for even a small amount, the guy next door (and/or your first client) can do it for free.

Also note that the GPL is perfectly fine with me improving a GPL-ed program in-house and then "not giving back". But as soon as I give or sell the improved version to someone else, I'm required to provide the sources to my changes to them. And they get the right to redistribute / resell / publish the whole thing.

Is this actually "left copyrighted"?

Hi again,

To provide the intent of the GPL specifically for MySQL, it goes like this... The database engine lives under the GPL, If I am a super genius coder that comes up with a new, and substantially improved sorting algorithm that I want to attach to the existing MySQL code, because the SQL interface is just what I need, then I am free to do that, and sell the new product. The agreement is the source for my improvement goes back in the pool.

If I make a killer database application with really clever inner and outer joins, as well as some super slick compiled lookup functions using MySQL, then I wish to sell the new application to any and all, I am free to do so, and should not have to pay a license fee, and also not be responsible to return my application to the community, because it uses MySQL, and is not part of MySQL.

The problem is deeper, however, because code that was added to MySQL to make improvements with isolation of users (limits users and adds security) and parts of the threading process, which according to the GPL should be made part of the open source, are being maintained as proprietary because it comes from the company's other application branch. Comes from engineers taking the best parts of two systems (because they had access) and creating a really excellent product, but it resulted in a Pandora's box situation.  Oracle is by far, one of the best data base systems available, but some of their business dealings cause me to be concerned with their intentions in the open source community.

Thanks Clem, and it was, yes "copyleft" is the term that I have heard.

Jack

Jack Chaney wrote:

The agreement is the source for my improvement goes back in the pool.

 

You should read the GPL carefully. You are under NO OBLIGATION at all to put the improvement back in the community. The INTENT of the GPL is that this will happen, but it is not worded that way. What the GPL requires is that if I sell someone the binary, I HAVE to provide them access to the sources (of the original AND the improvements).

If I want to make money by selling my small addition to a big GPLed project, I could try to do so, but anybody who buys my addition, has the right, according to the GPL, to become my competitor and sell my addition for half the price. They didn't have to write it so it is cheaper for them to sell it. This price battle should continue until someone legally puts my additions on the internet for free and distributes it to everyone who wants.

In practice, "putting it back into the pool" is what usually happens. I've written lots of improvements of GPLed programs because one thing or another bothered me. Because I don't want to have to patch those programs again next time I use them (say after a system-upgrade), I submit the patches back to the authors, in the hopes of them including it, and next OS version coming with those changes.

Example: My backup-machine was upgraded last friday and then it asked to reboot. I let it. Then it didn't come back up for a while. It's headless, so I couldn't see what it was doing. I guessed it was fsck-ing. A bad design-decision in fsck.ext3 causes the fsck on MY filesystem (in that machine) to take at least a month.... My fix was luckily incorporated into the upstream fsck sources and the fsck finished just as I managed to find a monitor.

The FSF is an organization that has an agenda. They occasionally interpret the GPL in a way that suits them.

In the original GPL, a "derived program" is like we're discussing here: An improvement that makes the whole program better. Or a different program that uses part of the original. But if some company make a package that includes say "linux" and then adds a few proprietary binaries required to provide the sources to those binaries? FSF has decided that pushing for that would not be in their interest. So they draw the line at linking with the GPL-ed sources. But this is an FSF interpretation of the GPL. Drawing the line there is "reasonable" and  this has resulted in nobody challenging that part of the GPL in court.

FSF may agree with your point-of-view that it would be best if enhancements are fed back into the ecosystem directly. However, we have to hold to the legal texts that govern the issue, in this case the GPL.

The GPL intent is to return fixes and enhancements to the community.  Like any contract or license, the recipient is free to ignore that element, because it is not clearly stated.  But the problem is, the community suffers as a result.

Jack

Right. The Intent is that modifications are given back to the community. To prevent lots of different forks each with their own enhancement.

But the intent is also to give FREEDOM. So for example I have the right to take the sources and study, modify and enhance them to my hearts content without giving back to the community if I so choose. It's just that I'm not allowed to redistribute binaries with those changes. This is one of the freedoms that the GPL gives me.

With respect to "giving back to the community", although that is the intent, the GPL was written at a time that "you have to put it on your website for free download for all" was not yet possible. So they formulated in such a way that if you distributed binaries to someone you have to EITHER include the sources, OR be willing to provide those sources on demand.

The intentions of the FSF/RMS are clear. They have written down rules by which to abide, which they thought would accomplish their goals.

Independent of what you or I think what their goals are, we have to follow the rules that have been written down. Even if you think that following some rules to the letter does not coincide with your intent, or what you think the FSF's intent was.

I seem to have lost something in translation...

I am not trying to assert the virtues of the GPL or open source (which is another of my quests), the point I am asserting is, when Oracle, purchased Sun, the listing of assets in the bill of sale included MySQL, which was, and should have remained owned by the open source community. Sun was never the "owner" of MySQL, but, only a supporter and contributor.  Because they didn't own it, there is no clear ability to transfer ownership to Oracle.

Jack

The thing about the GPL that I've only recently come to understand completely (maybe because of this thread, or something on hackaday), is that dual licencing is possible.

If I write a book (*), the normal procedure is that I own the copyright, I licence say penguin books to print the books provided they pass a commission to me. I allow Amazon to sell Kindle copies, provided they pass a commission to me. And maybe I set the condition that they ONLY sell the electronic kindle copies. The situation is that I own the copyright, copying is prohibited by default, and I can grant permission to make copies and set the conditions for that permission any way I want.

Next I write a program. I own the copyright. I can "grant permission to make copies" any way I like. I can for example, use the GPL to set conditions on the "making copies".

On the other hand, I could ALSO give permission to make copies to someone else, but set a different batch of conditions. This is the dual licencing that sometimes happens. And it is what Oracle offers to do.

That said, MySQL was originally written by MySLQ AB. They basically owned the copyright on anything they wrote. They got bought by Sun in 2008 and Sun got bought by Oracle in 2010. Those copyrights passed on to Oracle, in a perfectly legal way.

There is the question of patches and submissions. Has the community every provided patches that were included in the "mainline" sources? If so, the author of those patches by default has copyright on that portion. This would officially make it difficult to impossible for the "main" copyright holder to give out the "right to make copies" under a different licence. For the included patches they would have to get permission to do so from all patch-authors.

A way to prevent this situation is to ask the authors to sign over the copyright holdership to the main-copyright holder. This is what the FSF is doing. If I write a useful patch for "sort" (#), they want me to sign over the copyrights to them. This would allow them to legally licence my code to someone else under a non-GPL licence. For the Linux kernel, Linus Torvalds knowingly does NOT request patch authors to do so. It is now impossible to get permission from ALL 6000 contributors to licence the Linux kernel under a different licence.

If MySQL AB and Sun have always refused patches and fixed bugs in a "clean room" setup without looking at the submitted patches for problems.... then Oracle can legally relicence MySQL under a different (closed) licence. If in the history of MySQL patches from others have been included (as seems likely), then the authors of those patches can demand that Oracle NOT represent them in giving out a different licence to others. Or set conditions under which that licencing can happen. Maybe some money would need to change hands...

(*) This is made up. I don't write books.

(#) This is true. I write patches to sort.