Montag, Oktober 25, 2010

Opinion: The Cold War Between and LibreOffice - Linux Magazine Online

For the past month, I've wanted to express an opinion about LibreOffice and However, I've refrained, because I didn't know what to think either way. I still don't, but my conviction is growing into that this uncertainty is worth expressing. Although many people want to see a hero or a villain in events, I'm not convinced that anyone who is involved deserves the uncritical support of the community.

If you follow free software development at all, then you know what's been happening. Mistrusting Oracle's intentions towards, a group of developers announced The Document Foundation, an independent organization that is developing an fork called LibreOffice. In response, Oracle has declined the invitation to join The Document Foundation, and the Community Council has called on its members who are also members of The Document Foundation to resign to avoid a conflict of interest.

In the past week, these events have seen people on the Discuss mailing list choosing sides and justifying their reasons for doing so. Only the independent Documentation Project has announced a policy of neutrality, inviting writers "to work on other flavours of OOo, should they wish to do so."

Amid these events, I hesitate to take sides, because, so far, all we really have is rhetoric from both sides. The Document Foundation's home page sounds idealistic, but, then, so does Oracle's news release pledging continued support for Both sides have yet to back up the rhetoric with enough actions to make any serious judgment possible.

In this circumstance, I can only judge by past performance, which makes both sides seem less than ideal.

The Oracle back story

On the one hand, although Oracle has been involved in free software for over a decade, it has been unable to gain the community's trust. With its own projects, it has the reputation of being unwilling to relinquish control to the community, and of manipulating community expectations for its own purposes. -- of acting, in other words, how many people in the community are always afraid a large corporation would act.

For instance, while Innobase and Berkeley DB have continued to exist since purchased by Oracle, many suggest that development of both databases has slowed, to keep them from being a threat to Oracle's commercial products.

This is exactly the concern that led Michael (Monty) Widenius, the main writer of MySQL, to be concerned for its future when Oracle acquired MySQL along with with its takeover of Sun Microsystems. Although Oracle pledged to keep MySQL alive, Widenius believes that "They will market MySQL as an entry database that you can use until you afford something real," and phase it out after five years. He points to a halving of the Oracle staff working on MySQL as proof of his predictions.

Granted, does not compete directly with Oracle's products. Even its Base database does not, because it is a much more basic database. All the same, considering how Oracle has acted elsewhere, skepticism about its plans for seems well-founded. The fact that Oracle has yet to match its actions to its rhetoric suggests either it does not understand the community (and may therefore blunder in its future handling of or else it plans to act exactly as people fear (in which case the fear is completely rational).

The morphing of Go-OO into LibreOffice

On the other hand, The Document Foundation does not inspire total trust, either. Judging from the members, particularly founder Michael Meeks -- to say nothing of its motivations and goals -- The Document Foundation is simply Go-OO reborn.

Go-OO is (or was) a semi-independent group of developers, who became impatient with the slow pace of development under Sun's development. Several of its members, including Meek, are Novell employees, and the project has been funded by Novell. Since did not accept all its code, Go-OO quickly became the version of used by many distributions, including Debian and Ubuntu -- which is why Shuttleworth's instant support for The Document Foundation means nothing; in announcing that Ubuntu will ship with Libre Office, he is merely saying that Ubuntu will continue to act as it has in the past.

Over the years, Go-OO has, in fact, introduced enhancements faster than However, "faster" is a relative term, and The Document Foundation's rate of development is unlikely to be that much better than's. Similarly, whether Go-OO was a more democratic place for development is hard for an outsider to decide. Even if it was, will a relatively small organization be able to scale successfully to become an independent Foundation? Although Go-OO has made no obvious blunders, it seems only a modest success.

Moreover, Richard Stallman's words of welcome when LibreOffice was announced strike me as ironic. Less than two years ago, other free software supporters (although not Stallman himself, so far as I can tell) were condemning Go-OO as a tool of Novell and the conspiracy to spread the use of Mono.

Now that the cards have been reshuffled, and Oracle is replacing Novell as the company that parts of the community love to hate, The Document Foundation looks better than Go-OO ever did. Yet there is no reason to think that the goals have changed with the name. If you believed that Go-OO has a hidden agenda, then you can believe with equal justification that The Document Foundation has one, too.

A Bluff That Was Called

What happened, I suspect, was that Go-OO, already chafing under Sun's tight control of's direction, saw more of the same -- if not worse -- awaiting in Oracle. Hoping to succeed before Oracle could articulate its plans, Go-OO members reinvented themselves, and announced the foundation that they had long been calling for. But Oracle refused to be stampeded, and escalated the fork into something that resembles corporate warfare.

Whatever the merits of either side (and I am most inclined to support The Document Foundation, although only on the principle that any number is greater than zero), I suspect that the losers in this situation will be the users. The risk is that time will continue to be spent in flame wars that could be better spent in coding. What seems likely is not only a general division and duplication of effort, but, in Oracle's case, a decision to focus on proprietary development as a defensive measure. By making the gambit that it did, The Document Foundation may have perpetuated another version of the stalemate that it was trying to break.


Dienstag, Oktober 19, 2010

Is it a spoon? Is it a fork? No, it's LibreOffice!

The newly founded Document Foundation announced their existence on Tue, 28 Sep 2010. Their first product is LibreOffice, a version of, which is released under the LPGL v3+ license.

What are the reasons and -- heatedly discussed -- is this a fork or not? Let me discuss some of the background and give my own layman interpretation of what is happening. Even back in 2008 Michael Meeks was looking at the contributor statistics, finding that Sun-external contributions were comparatively small. In the graph, you see around 30 active Sun Microsystem contributors, and between 4-8 external contributors. That can be interpreted as a sign of tremendous commitment by Sun (which has been significant all the time), or as a failure to mobilize external contributors. After all, this is 4-8 active external contributors when the Office Suite is the killer app next to a web browser and email application for most office workers! I have not checked the development in 2009/2010, but I don't expect that things have looked much different.

Micheal Meeks and others blame that on "enormous groupthink", but also on the fact that Sun, and now Oracle, requires Copyright Assignment for all code contributions, so they can sell proprietary versions of the office suite, while others cannot. This prevents some companies employees from contributing and is a disincentive for many individuals too. (e.g. University employee's code often is automatically copyright to the employer, and they cannot transfer their copyright). Anyway, the end result is that Sun, now Oracle, never played on one level with other contributors, which is what they had promised from the very beginning.

In their original press release on July 2000 , Sun had stated:

[...] In addition, Sun also announced today the new Foundation, which will initially be modeled on other successful open source projects and will consist of a project management committee, source code maintainers, and developers. Sun will hold a equal membership position in the Foundation project management committee. [...]

That foundation has failed to materialize until today, and so has the equal membership position. As I am not familiar with the core (people and processes) of the Documentfoundation, I can't comment on how it came about, but they announced their existence, emphasizing that they see themselves as part of the community, implementing the decade-old promise of Sun of a level playing field, doing away with Copyright assignments, and at the same time, lowering the barrier of entry for prospective developers.

Given the surprise that many showed, the process seems to not have been handled very transparently. Oracle was "invited" to join the new foundation and help to shape it, ideally donating the trademark to the foundation, however it seems that there was little discussion going on informing Oracle about specific plans beforehand.

So far so good, the have been 3 (related) contagious issues which are heatedly discussed:

1. Is LibreOffice a fork or rather than the natural further-development of OO.o? As in, the same thing, the same community, with just a different name?
2. Does the OO.o community also represent LibreOffice users, ie can someone be on the bouard of the OO.o community council and at the same time be in the board of the Document Foundation?
3. This leads to a bad fragmentation of the Office suit developers and should be avoided at any price.

As somewhat sensationally reported on Slashdot does the Community Council of the project expect people (lengthy IRC meeting minutes here) with a role in Libreoffice to step down due to conflict of interests. Oracle proponents state, that there obviously is a conflict of interest, as the 2 projects were effectively competing now. Most TDF proponents were trying hard to emphasize the fact that the project were not competing, and Oracle were still invited to join, and that there is no conflict of interest as both organizations goals were congruent.

So what is my take on it?

1a. It is a fork, let's get over it. The organization's goals are the same, and the software might be the same, but the means to achieve that are very much different. If this is not a fork than gcc/egcs, xf86/xorg, emacs/xemacs were no forks either. The "but we invited Oracle to join" argument doesn't count, gcc adopted back the egcs code base later as well, and it still was a proper fork. So even if Oracle joins in the fun and starts using LibreOffice as code base, this would have been a fork.

Was this fork necessary? In my opinion yes. As former Chief Open Source Officer at Sun Microsoft put it: "If the company sponsor stands still and the community moves on, who forked?" I sincerely think that Sun/Oracle is holding back the potential that the Office Suite could have. I have the highest respect for those currently involved on the Oracle side of things and I don't think any of them is at fault, but the current OO.o project governance structure is killing of external contributions. Leading to the situation where Oracle has to contribute nearly all of the source code.

Does this mean there can be no cooperation? Heck no, "competing" implementations inspire each other in the FLOSS world all the time, and both project rally around the .odf format. So there is plenty of room to collaborate anyway.

1b. There have been arguments that this belittles Sun/Oracle as a contributor, showing disrespect of the enormous amount of code they put into the code base. The code is certainly huge. Including comments about 8m lines of C++ code and 410k lines of code of Java plus various other stufff sprinkled in. This compares roughly to the whole of the linux kernel which wheighs in at approximately 13m lines of C code (for 2.6.35). But this is a two-sided sword: If Sun were more open, they propbably would not have to contribute all the code themselves. This is a huge project, and having to bear it on ones organization shoulders' is an incredible task, especially when directly and openly competing with Microsoft's office cash cow.

2. As for the conflict of interest, this depends. A conflict with the OO.o community council? Certainly, now that the three indedependent members have been asked to step down, the council will be 100% composed of Oracle employees. That doesn't make these employees and contributors bad or ruthless, but it makes it an enlarged-arm of Oracle, rather than representing the community. There is effectively no independent user community anymore. And with that council, an independent member is certainly at a conflict of interest. Are there conflict or interests with things like I don't think so. The code base is still mostly the same, the UI is still mostly the same, and the templates can be used for both office suites. There should be plenty of collaboration and dual board-membership between those sites. (Dual board memberships are quite common in the business world).

3. Does this lead to Office developer fragmentation? It is true, that the amount of direct code reuse will probably decrease over time as the code bases diverge. But at the same time the current governance style prevented many contributions from being taken, and everyone had to maintain their own little stack of patches, leading to increased fragmentation. Redhat, Debian, all had maintained and applied their own set of patches on top of OO.o. It is also a well-hidden secret that most people using on their Linux-desktops were not really using but rather (wikipedia entry) which collected up to 800 patches on top of vanilla Distributions, such as Debian, Mandriva, openSUSE, Gentoo and Ubuntu have been using some or all of the Go-oo patches for quite some releases. Being able to merge these patches into the proper code base and also integrate those of, say, Redhat, actually might lead to less fragmented development that it had been before.

I wish LibreOffice all the best, and do hope that Oracle joins in the fun as one among equal as the year 2000 press release had promised. The intransparent process and the hot blood that had been shed in the mean time might make that much more difficult and the blame does not squarely lie on one side or the other here, but it would be the logical and correct thing, in my private opinion. Look at Eclipse and the Eclipse foundation for one example how things might work out in the long run. Oracle should get credit for what they have contributed to the free and Open Source world.