Sun Refuses LGPL for OpenOffice; Novell forks

TRS-80 writes "Kohei Yoshida wrote a long post on the history of Calc Solver, an optimization solver module for the Calc component of OpenOffice.org. After three years of jumping through Sun's hoops on his own time, Sun says it will duplicate the work because Kohei doesn't want to sign over ownership of the code. Adding insult to injury, Sun then invites him join this duplication. Because of Sun's refusal to accept LPGL extensions in the upstream code, Michael Meeks (who recently talked about Sun's OO.o community failings, and ODF and OOXML) has announced ooo-build (previously just for build fixes) is now a formal fork of OpenOffice to be located at http://go-oo.org/. "

Not an official "Fork"

[mgpeter]

This is not an official "fork" of OpenOffice.org. This is simply a way for Windows user's to get a nice "Development version" of the office suite similar to what is deployed on most GNU/Linux Distributions. Of course if you don't want to use a "Development Version" on your workstations, you can get a stable version of the OOO-Build service with Novell's version of OpenOffice.org for Windows (which is what I prefer).

Re:Conspiracy theory - MS behind all this?

[morgan_greywolf]

The OOo community, or, that is, both developers not working for Sun ;), have been talking about a fork for a while, well before the whole MS/Novell thing. Sun won't take any code from anyone not willing to assign the copyright to them, which pisses a lot of people off. They also won't take code that deviates from the strategic direction Sun wants to follow. Development proceeds at Sun's pace, which as some say -- giant land tortoises move faster. The OOo community needs to fork OOo for the good of the project.

Why demand signed-over ownership?

[Bert64]

Why do Sun demand that ownership is signed over, can't they just accept dual licensing - that is you license it under the LGPL and license it specifically to Sun under other terms (eg BSD) so they can reuse it in staroffice.

Re:When will people learn?

[Shawn is an Asshole]

While I agree with you that forking generally isn't good, at times it can possibly be a good thing. Take a look at XFree86/Xorg. Since the fork Xorg has had massive improvements, finally getting X to a modern state. Hopefully this fork will work on improving OOo, specifically in the GUI and speed areas (Novell, please at least copy Lotus Symphony's GUI or MS Office 2004 (OS X) but implement in native controls making use of system settings (it should follow my icon theme and font settings at least)). While I use OOo, it really doesn't seem as if Sun has much of a goal for it. The GUI isn't very intuitive, it still is horribly bloated, and overall it doesn't integrate with the system and looks hideous. Each new release doesn't seem to have any noticeable improvements over the previous. It just feels really stagnant. Hopefully this fork will have some direction and actually have a goal of competing with MS Office.

They really need a goal like this [launchpad.net].

Re:When will people learn?

[Anonymous Coward]

Hopefully they rip all the Java out too.

Coding is commodity

[XMLsucks]

When you contribute open source code on your own time, it is an implicit admission that your code is worth little, and so don't be surprised to see someone else take the same view and duplicate it! The value is the fun in writing it, thus there will be some handful of people on the planet that share the same sense of fun, and will duplicate the work. I've seen lots of my stuff duplicated. And I've duplicated other projects. That is how people have fun and learn.

Imagine if you'd gotten money from Sun for your code. Would you care (as much) if they ignored the code? They'd have the right by having purchased it. But having spent money on it, they'd probably be less likely to discard it, and to start from scratch. Money makes a difference.

Jeez, this post is the typical complaint seen in charity work: "Oh, they didn't value my work, and I have no sense of self-worth, so now I'm all upset!" "The people running the charity are all in a clique and don't pay attention to the contributions of the other charity workers. They're destroying the spirit of the organization. Lets go create another organization that cares!" And then the cycle continues. The basic mistake is in thinking that other people have to value your work. They don't. Only you do.

Re:When will people learn?

[femtoguy]

I don't know that his is actually so bad. Remember that open source is all about choice. In the proprietary world, there is a huge advantage to being the one standard program, and so companies have used file formats to guarantee their positions. In the open source world, the open office xml file format is an open standard that anybody can use. We can easily have IBM with their office suite, Sun's Star Office, OpenOffice.org and a fork of it, KOffice and everybody can choose whichever version they want, as long as they use the standard file format. It's perfectly analogous to the web. It doesn't matter that some people use IE, others firefox, and others iCal or lynx, because html is standard, and anybody can implement it. In the end it is data that matters, not programs or platforms. This is the great strength of open formats and open source. Let people choose their programs based on their features and use interaction rather than being forced by format externalities.

JCA

[mmurphy000]

For core changes to the OpenOffice.org code base, Sun requires joint copyright assignment (JCA), whereby both the original author(s) and Sun jointly hold copyright. This allows Sun to relicense the OpenOffice.org code as needed (e.g., GPLv3).

IANAL, but with the JCA, nothing would prevent Kohei from making his code available under LGPL or any license he chooses outside of OpenOffice.org. However, by not signing the JCA, Kohei is preventing his code from being part of the core Oo.org code base. For whatever reason, the Oo.org team must want a solver that is part of the Oo.org code base, so if Kohei won't sign the JCA, there are few available options.

What would be interesting is if there were a way to basically split Kohei's solver component into three pieces. One is the GUI layer (there's menu choices, presumably leading to solver-specific dialog boxes), one is the bridge to communicate with the underlying spreadsheet data, and one implements the solver logic proper. Packaging that last piece as a LGPL third-party component, reusable among other projects (e.g., Gnumeric), might be acceptable to the Oo.org team, provided that the Oo.org-specific UI and data access bridges were part of the core project. I have no idea if this kind of code split makes any sense, since I've never written a solver, though Kohei references lp-solve, suggesting that part of his code might be able to be split into an nlp-solve...

The power of Open Source

[speaker of the truth]

Only in open sourced code could a fork like this be made. If it had been Excel he had written this code for he'd probably be getting sued for breeching some patents.

Re:I'm getting this feedback often...

[jsight]

3. The ODF spec is overly long and needlessly complex, to be implemented faithfully.

I was with you up until this point. People who think this spec is long don't realize just how complex this stuff is. If you want interoperability to actual work, the spec needs to be much more comprehensive than the ODF spec actually is.

Have you noticed that the .ods spec doesn't even provide a comprehensive list of supported functions?

It's not needlessly long, its too short.

Re:When will people learn?

[suv4x4]

The GUI isn't very intuitive, it still is horribly bloated, and overall it doesn't integrate with the system and looks hideous.

Hmm it wasn't long ago I heard praises of OO since while Office 2007 changed its UI dramatically to deal with control bloat, OO kept the 2003-style interface. I mean you do realize: Open Office literally has the Office pre-2007 UI, in fact OO has less controls and toolbars than Office 2003 did.

I'm seeing more and more opinions in the other direction, which means the tide is turning. I guess the infamous Ribon wasn't that bad after all.

Re:When will people learn?

[Anonymous Coward]

Scott, the solution would be to share the copyrights with a non-profit that would have some kind of process that would make relicensing quite difficult, even though not impossible. This legal entity could be defending the code. But creation of this kind of entity would be contrary to the will Sun has to control everything. I am currently maintaining a WordPerfect import filter. A code where Sun did not do a single line if not some build breakages fixes when their build-system changed. I stupidly assigned once upon the time my copyrights to this code to Sun (as other of the authors did) in the hope that "OpenOffice.org community" will help to maintain it. The only error we made was to believe that there is something like a developer community. Now, Sun owns the code where they did not do a single comma. And the only contribution was to complicate the updates so much that one gets simply discouraged

Re:Coding is commodity

[JohnFluxx]

I also look after my wife and children in my own time. Is this an implicit admission that they are worth little as well?

 

No! One missing person does not mean you lose!

[jimwelch]

I am NOT a lawyer, but my father-in-law is! (hmmm, so what?) As some recent cases in Europe show, They may "steal" GPL code from many people, but if one copyright holder sues and wins, everyone wins! So the assignment clause, does not hold water on this ONE argument. It does server other purposes and is a valid requirement, but comes with a negative side too. I personally will not contribute to any project that has this requirement. My contributions, have stayed my own, under GPLv2.

That being said. I still think this fork is *long over due* and is required to "patch around damage". Sun has made too many mistakes and too many enemies.

The other example of a good fork, is Joomla!

Not quite what I hoped for

[Anonymous Coward]

Ever since Sun shoved Java into OOo, didn't use SQLite for Base and continues to stick to that horrible interface, I have been hoping for a fork. So kudo's to the free software world for that.

But... it's Novell, people. Novell. Might as well be MS. When will we see a real community-run/owned fork?

The FSF reasons for signing over ownership

[Per Abrahamsen]

The FSF requires assignment of ownership for "core" components such a GCC. There are two reasons for this:

1) It is (legally) easier defend the license if ownership is clearly defined (and before you comment: The law is rarely Boolean).

2) To make it possible to re-release under different licenses.

The GPL2 to GPL3 is a poor example of #2 as they usually add a "any later version" for their GPL'ed source. But ownership gives them the right to give permission for other free software projects to use FSF code in projects that use other licenses, they are quite pragmatic with regard to such licenses.

Both should paply to Sun as well, plus the added ability to make proprietary versions (like StarOffice) which may link to other peoples non-LGPL compatible code.

ooo-build has long been more than build fixes

[soullessbastard]

Disclaimer: I am a founder of the NeoOffice [neooffice.org] project.

ooo-build has long been much more than build fixes. For many years it has been the public face of the work Ximian and Novell have poured into the OpenOffice.org source base. It has a long history of features that Ximian/Novell have helped develop, including (but not limited to):

• OpenXML import/export support via odf-converter
• Kohei's solver optimization extension
• Native widget framework and GNOME integration (from back in 1.1.x)
• Visual Basic suport for Calc
• Alpha-blending and enhanced alpha blended icons
• A redesigned GNOME-like icon set
• Microsoft Works importer
• Evolution integration
• And more...

ooo-build is about functionality and features. Despite the name, it has never been about "build fixes" as indicated in the article. The additional functionality is so awesome that, at NeoOffice, we have been using ooo-build in NeoOffice [neooffice.org] since March and have been donating back bug fixes and Mac-specific support patches to the ooo-build project. Years ago the Ximian work on OOo 1.0.3 was so promising that I put together a Mac OS X port back in 2003 [neooffice.org] which folks used for a long time. OxygenOffice [sourceforge.net] also is based off of the ooo-build project (although I do not know if the OOOP team coordinates with ooo-build).

The ooo-build team has done amazing work. It is sad to see their work go unrecognized by so many and be outright rejected or stalled by Sun. NeoOffice users have loved having the functionality ooo-build brings currently and continues to bring in the future, and much of the work pioneered by ooo-build is critical to maintaining the Mac platform as a viable office solution (read VBA). Sun's lack of acknowledgement and incorporation of ooo-build features does nothing but hurt users. Having received a "you're welcome to join us" response similar to Kohei [kohei.us], I am glad I do not consider myself part of OOo any longer. The freedom of forking has allowed NeoOffice to incorporate all good code without all of these politics and marketing games. Forking has allowed NeoOffice to deliver to Mac users the features they wanted yesterday regardless of where those features came from. Sun has a history of a "not invented here" syndrome at times when it comes to code within their "open" source projects.

I'm glad to see that ooo-build is getting some recognition. I hope more users start seeing some of the great functionality they can get today on Windows and Linux, and once again I thank ooo-build, Ximian, and Novell for their continued dedication to improving OOo.

ed

Re:Let that be a lesson

[Dunbal]

Different users want different things. If you end up with "One Big App" that tries to do everything anyone could ever dream of, you'll have all the creative, hard to find bugs, flaws and design problems that go along with such complexity. Which is where we're at today with slow, bloated, buggy programs.

      Remember that the whole POINT of software is to specialize. I'd rather see many forks of smaller, specialized, GOOD apps than the "One App to Rule Them All" approach. So long as we keep track of all the different versions and where they particularly shine and figure out a way to search for the "right version for me", forks should not be a problem but a BONUS.

take notice: Java

[m2943]

Now would be a terrible time to stop developing parallel languages, because the problem is just now coming to the forefront with the limits of single-core performance pushing back and multi-cores taking over.

This is one of the reasons dual-licensing is bad. Big projects with this problem are OpenOffice, Java, and Qt.

ooo-build (previously just for build fixes) is now a formal fork of OpenOffice [CC] to be located at http://go-oo.org/ [go-oo.org] [CC]

And this is the proper response: to fork the code and make an open-source only version, leaving the company and all its legal shenanigans in the dust.