The Decline and Fall of OpenOffice.org
Off the Beat: Bruce Byfield's Blog
When LibreOffice first forked from OpenOffice.org, I was unsure how to respond to it. Many of its founders were members of Go-OO, the sometimes controversial not-quite-fork, so LibreOffice looked like simply a continuation of GO-OO under another name. However, since last week, when Oracle donated OpenOffice.org to the Apache Software Foundation, LibreOffice.org is looking more and more like the natural heir of the original OpenOffice.org -- by which I mean the project that will do the free software community the most benefit..
Not that there's no room for two projects working with the same code. And, as Allen Pulsifer suggested, perhaps LibreOffice supporters should to join the new OpenOffice.org to keep track of what it is doing. For most people, the division between the two projects has always been more about governance than personal animosities, so coexistence is almost certainly possible.In fact, Apache representatives have suggested repeatedly that nothing would please them better.
The trouble is, I wonder whether the Apache Foundation is really that enthused about its new proposed incubator project. True, Apache members are methodically going through the motions of setting up the project; after all, they've started dozens of projects in the past. Yet the tone on the mailing lists and public statements suggest to me that Apache members are generally non-plussed about the donation, as though it is something they are dutifully accepting rather than anything they're excited about. Alternatively, perhaps people are restraining themselves because of behind-the-scenes discussions that haven't finished yet.
Either way, so far the new project is small. Doing some simple arithmetic based on the LibreOffice Credits page, I estimate that some 220 people have committed patches to OpenOffice.org in the past, who have not committed any to LibreOffice. So far, though, only 83 initial committers are listed on the incubator wiki. Only 22 indicate previous involvement with OpenOffice.org, either through their email addresses or their affiliation. Of the rest, 6 are connected to Apache, and 1 to Oracle. Another 6 are connected to IBM, which develops the Lotus Symphony fork, and 4 to RedOffice, a Chinese OpenOffice.org fork. This a small number for a code base as large as OpenOffice.org's, especially when less than half appear to have worked with the code before. Apparently, contributors are not exactly flocking to the new project.
From what I am told, the new project's acceptance as an incubator project is almost certain, and so is its eventual acceptance as an Apache Foundation project. But even if all goes well, organization and governance is going to be a distraction for some time. Add the small number of contributors, and the new OpenOffice.org seems unlikely to release a new software version for at least four to six months -- possibly, for as long as a year.
Meanwhile, LibreOffice is developing rapidly, producing code dual-licensed under the Mozilla Public License and the third version of the GNU Lesser General Public Licenses (LGPLv3), which the Apache Foundation regards as incompatible with the Apache License.
By contrast, code released under the Apache License is regarded as compatible with both the Mozilla Public License and LGPLv3.
In other words, LibreOffice will be both months ahead of OpenOffice.org, and able to borrow OpenOffice.org code, and OpenOffice.org behind and unable to borrow LibreOffice code.
Under these circumstances, how can OpenOffice.org ever hope to compete? Unless something unforeseen happens, OpenOffice.org will start from behind, and never have a chance of catching up.
It doesn't help, either, that placing OpenOffice.org under the permissive Apache license is yet another barrier to attracting contributors. Free software contributors will undoubtedly prefer a copyleft license that requires contributions to be released under a similar license. The donation to Apache is widely seen as interference by IBM, which is believed in the community to want the ability to sell proprietary versions of the code as Lotus Symphony.
For many, this abandonment of copyleft is yet another reason for staying away from OpenOffice.org. With the Free Software Foundation endorsing LibreOffice over OpenOffice.org for precisely this reason, licensing is still another barrier against OpenOffice.org's success.
The LibreOffice Advantage
Despite the difficulties that the new OpenOffice.org faces, some people favor it as a more stable home for the code than LibreOffice. Their assumption is often that Libreoffice is a disorganized, small project.
The truth is, it is neither. Far from being disorganized, LibreOffice -- or rather, The Document Foundation that develops it -- is registered as a German non-profit under the name of Freies Office Deutschland e.V, with the long-established Software in the Public Interest (SPI) representing it the United States. Although it has yet to democratically elect a board, The Document Foundation has managed a successful fund-raising campaign without interfering with organization or development.
Nor is it a small project. From the figures on the LibreOffice Credit page, I estimate that it has at least 367 members -- four times the numbers that OpenOffice.org has managed to attract so far. True, only about a third of that number contributed to OpenOffice.org before the fork. But it has managed to attract some 243 new active contributors in the nine months of its existence, more than replacing the 220 contributors to the old OpenOffice.org who have not worked on LibreOffice.
In contrast to OpenOffice.org supporters, the LibreOffice community seems active and engaged in what it is doing. Although the mailing lists are more sedate now than when they were first set up, the LibreOffice community conveys a sense that everything is up for reconsideration -- and that everyone is eager to implement changes. When LibreOffice and OpenOffice.org each released new versions in February 2011, LibreOffice was already the most innovative, and all indications are that it will continue to improve far more rapidly than OpenOffice.org under Apache can ever hope to.
I have nothing against Apache's version of OpenOffice.org -- even though the donation does seem like one last spiteful gesture by Oracle against The Document Foundation. But given the challenges it faces and LibreOffice's head start, I also see little chance of it succeeding
The new incubator project may have the name and the trademark, but it appears to have no future. At best, it seems doomed to hobble along for a few years, falling farther and farther behind LibreOffice until it collapses from irrelevance and neglect.
Disagree with Grey GeekI think you are mistaken on the BSD license being a hindrance to *BSD projects. You don't seem to be aware of the USL lawsuit, which devastated the credibility of all BSD projects for several years, while all were niche projects. Prior to the lawsuit, they were perceived as being the premier open source operating systems.
At the same time, Linux was still pretty small too, but it was starting to grow, and the lawsuit very much benefited Linux.
I personally believe that licenses have very little to do with success of a project, since most developers don't care what they are, and in fact the GPL had negatives to it for Linux -- perceived closeness to GNU, association with fringe political advocacy, and the fact that it's restrictive, even if that means it's for your "freedom".
A lot of people say that the BSD license deters developers. As far as the appeal, I strongly disagree. If you mean in terms of developers who simply put their energy into proprietary forks, then I can see that, however, in many cases, the BSD projects have receieved major contributions from third parties. Some examples include the LLVM compiler suite, launchd init system, Solaris ZFS, Webkit, and many others. Many of these third party programs were tremendously lacking prior to their introduction, in particular the compiler kit. ZFS has become a high priority for FreeBSD developers to lure server users, and is under a license that can be linked with the BSD license, whereas the GPL is more demaning in it's requirements, and cannot be directly used with it.
You also assume that every BSD licensed project has had dozens of significant forks, but in many cases this is not the case. Regardless of forks, does it really matter? I don't think it does. I would be hard pressed to think of many projects as successful as Apache, perhaps you disagree. Some developers may refuse to work on BSD licensed projects, whereas for perfectionist reasons I avoid developing on copylefted projects whenever possible. (I want to contribute towards public domain, MIT/X, BSD, etc...) I would also wager that X would have died in the early 90's, were it not for it's liberal licensing. You might point to proprietary forks, but I would point that it would have been unable to maintain influence over technically superior windowing systems such as NeWS, without it's liberal license.
If BSD has a bad survivial characteristic, it is that they want to support existing technologies rather than create their own, which creates a perception of stagnance. It is definitely not their choice of license. (I'm not saying that the GPL would harm a project, just that not every project needs to be copy left)
Part of the Apache OpenOfice story everyone seems to be missingIn all of the debate on whether LibreOffice would have been a better home than Apache, most people are missing a very big part of the story. The document space is poised for a rethinking. Documents today are relatively unchanged from what they were many many years ago. They are fundamentally constructed of 2 layers, a content layer and a presentation layer. Documents are going to evolve. They are going to become a substrate for new collaboration models. They are going to get a 3rd rich semantic layer. They are going to become much smarter. As we increasingly assemble information from fragments issues of trustworthiness, provenance, navigation and discovery, policy and regulatory compliance, data leak protection and social magnification become increasingly important. In this new world significant value flows to advanced enterprise content management, and business, social, and predictive analytics. They become part of the knowledge basis of a company and an essential component of reputation. Can you imagine the potential of a Watson QA machine that can chew on semantically rich documents to ascertain relevance, timeliness, reliability, accuracy...? So quite beyond the "who can create the best word processor, spreadsheet and presentation editor?" question... we should be asking where will the source of innovation come from. What entities have the resources, intellectual property, expertise and will/interest to engage in a collective investment model to create new value on a large scale. IBM is one such entity but there are many others. Once Apache OpenOffice gets its feet under itself, one the the real measurements of its success will be which new players come to the table and what new directions they collectively invest in. I think the assertion by some that a non copyleft license will somehow inhibit development is mistaken. I think there are plenty of examples where companies and individuals come together to make collective investments because they understand that they all get to participate in value creation and magnification by doing so. The Apache foundation and the Apache licensing regime may very well be a much more hospitable, and yes stable/predictable, environment for critical investment and development resources that are currently "off the table". I for one hope that Apache OpenOffice and LibreOffice cooperate as much as possible. They should see themselves and complimentary and mutually interdependent. Hopefully a lot more value gets created that flows to customers and end users. OpenOffice, and ODF in a related way, is a pragmatic starting off point for a renaissance of document technology and innovation around rich semantics. I really hope, as would I think the LibreOffice supporters would as well, that it is wildly successful.
The numbers may be correct, but their interpretation isn'tI doubt that the future is predestined already.
The number of contributors to OOo that are familiar with the code may be small compared with the current Oracle staff, but it is bigger that what Libre Office had when it started. And I assume that the RedOffice and Symphony developers also have at least a few years of experience with the OOo code.
It is true that theoretically the LO developers can take code from the Apache OOo project, but only to a certain degree. Merging code from OOo already kept them busy for weeks; with more and more diverging code bases it will become close to impossible to continue that. So at the end we will see two projects developing independently, and their success will be defined by their own developers alone.
Besides that: did you ever think about what might happen if Attachmate lost interest in SUSE? This will remove most of the LO developers that are familiar with the code.
Re: Correctiondharbison says: "What party benefits from continued dissent here?"
A better question would be which party benefits from ignoring all the noise and BS and getting on with the job of improving the OOo code base - that would be the Document Foundation, who is the only group actually improving code right now. Apache/Oracle/IBM sure is making a lot of noise for a group that hasn't released any software in five months, and which looks highly unlikely to release anything anytime soon. Remember the 3.4 release that was planned? By the time we see OOo version 3.4 (if ever), LO will likely be on version 3.6, 3.7, 4.0 - who knows?
Lotus SymphonyI think dpharbison mis-read the article. I can't see anywhere that it states IBM is selling Symphony. It merely said that IBM is, "believed in the community to _want_ the ability to sell proprietary versions of the code as Lotus Symphony." I've no doubt that there are indeed those who believe that, and perhaps with good reason.
Measuring SuccessHow do you measure success? Why do you say the Apache project must "beat" LibreOffice?
It is nothing of the kind. Apache's version of OpenOffice.org must simply be produced to be successful. Having a permissively-licensed version of OOo is just as important as a Free Software version of it (LibreOffice).
I do agree that LibreOffice is in a position to remain "ahead" due to the potential flow of code, based on licensing concerns.
Yet I disagree with your characterization of the people involved. You fail to count all the people behind the Apache Software Foundation itself. There are dozens and dozens of people ensuring the Foundation provides support for: Infrastructure, Legal, Governance, Publicity and Marketing, Branding, Conferences, and more. The LibreOffice community is *beginning* to set up many of those things, being only eight months old. The ASF has been at it for over a decade.
The 80-something people on the initial committer list does not include those under a "wait and see" position. Give the project a year, and a release, and count again.
LibreOfficeWith a mere 1.3 million downloads in 6 months (http://listarchives.documen...teering-discuss/msg00848.html), how is LO any kind of success? OO.o 3.x had 100 million downloads in 1 year.
The fact Apache OO.o already attracted a QUARTER of LO's contributors in the past 6 months, in mere days, before even becoming an official project, including some original OO.o contributors which didn't join LO, seem to paint OO.o's future as brighter than ever!
LibreOffice Name a LiabilityThe biggest problem I see with LibreOffice in the US, is the name. The connotation of Libre here is negative and elitist. While the name OpenOffice is somewhat recognized, the response to LibreOffice that I've experienced is one of disdain, with most discarding even consideration based ONLY on the name.
Without support from at least some commercially recognized names, LibreOffice is pretty much a non-player in the US.
CorrectionIBM does not sell Symphony as you incorrectly state. Symphony is available at no charge, and as everyone knows is based on OpenOffice Technology. You may want to observe and report on how IBM does or does not contribute to the Apache OpenOffice.org project as it progresses.
You may also want to note that the Apache project proposal to start formal incubation of OpenOffice.org passed today by a wide margin, with vigorous Apache membership support. See http://bit.ly/kvwpDi ...again you made incorrect predictions.
Such is the state of industry journalissm today. Rather than sow seeds of dissent, it would be helpful to do the reverse. Don't you think?
What party benefits from continued dissent here?
RedOfficeExcellent article. The fact that RedOffice has 4 developers on Apache's OpenOffice (compared to IBM's 6), demonstrates they are the second commercial entity (and the silent partner) after IBM to benefit from Oracle's move of OO to Apache. I don't recall any RedOffice people being on connected to LibreOffice (or contributing code to it). Can anyone confirm or disprove my assertion?
Confusing description of the license comparisonsBruce, thanks for an excellent analysis of the OOo vs LibraOO situation. I was confused about one point in your discussion: the difference between their respective licenses and how that rendered them incompatible. But, one of your links cleared up the confusion:
<i>All distributions from the Apache Software Foundation are made under the terms of the Apache License, Version 2. This includes contributions made via mailing lists and issue trackers. All committers are required to sign an ICLA. ASF distributions can not include material made available under LGPLv3. Unmodified MPL libraries may be included in binary form. <b>The Apache License is (one-way) compatible with LGPLv3. </b></i>
The last sentence in that snippet explains why OOo and LibraOO source are not compatible and why, not being copy left, OOo is destined to fade while LibraOO takes over that niche in Linux AND Windows. It also explains why Linux is so popular and has generated so many applications, while BSD remains marginal and needs a Linux compatibility layer to gain access to the many Linux applications that are not available in a pure BSD environment. It is precisely because so many commercial software houses use BSD code and rarely return anything back to the BSD code base that BSD needs such a Linux compatibility layer, in effect sponging off of the GPL. Even more amazing is that those who plunder the BSD code base are only required to include an innocuous license file "somewhere" in their package, which usually ends up being installed where users rarely look, so the BSD developers rarely receive notice for their work.
DARPA and NICTA release the code for the ultra-secure microkernel system used in aerial drones.
Should you trust an online service to store your online passwords?
New B+ board lets you build cool things without the complication of a powered USB hub.
Redmond rushes in to root out alleged malware haven.
New initiative will bring futuristic virtual reality effects to the web surfing experience.
Dyreza malware launches a man-in-the-middle attack that compromises SSL.
New cloud combines worldwide access with local attention to data security.
A first cousin of the recent Heartbleed attack affects EAP-based wireless and peer-to-peer authentication.
FOSS community acts to protect freedom of choice for laptop devices.