Home » Uncategorized » Mike Conlon on the Apache OpenOffice fork

Mike Conlon on the Apache OpenOffice fork

The comment below was made on my blog, but it was so good, I’ve turned it into a post. It was written by Mike Conlon, who has written papers about forks in software.

It’s obvious that many of your correspondents either haven’t read your article [about the reasons against the Apache OpenOffice fork] or don’t understand it. However, I want to comment on forking.

First of all, the ability to fork is good. However, not all forks are good. Forking occurs when one set of parties in a development project develops serious objections to the direction the project is going. Once the fork occurs, usually the vast majority of the community choose one tine of the fork, and the other one withers.

Example: Remember XFree86? The leadership of the project became very unresponsive to community requests, to the point that many community members were complaining, and new contributors felt unwelcome. It actually forked into several tines, but soon most of the community settled on the X.org fork, which has thrived, while XFree86 went nowhere after the fork.

Sun was rather unresponsive to the OpenOffice.org community. Many potential contributors wouldn’t join the project because they would have been required to deed ownership of their code to Sun. But for whatever reason or reasons, the OpenOffice.org development community never grew to anything like the size of other open-source projects of similar scope. Then Oracle bought Sun, and many in the community simply didn’t trust Oracle to lead an open-source project properly. Sun was bad. Oracle was intolerable. The result was the LibreOffice fork.

LibreOffice appears to be thriving, while there seems to be little happening in Apache OpenOffice.org, other than press releases. This is a sign of a good fork. As some have commented, There is some instability in recent releases, but I also see the RIGHT THING™ where I didn’t see it before. (E.g., when you click on the change-the-font-size drop-down list, the current font size is now in the center of the list rather than at the top as before.) The increased pace of improvement is a major sign of the successful tine of the fork.

The Apache OpenOffice.org effort must be considered a fork of LibreOffice, even though it has its predecessor’s name. As you have indicated, the original OpenOffice.org project is dead. (The project is dead, but the Website lingers on…) So this isn’t the original, this is a fork. It’s a bad fork: it’s bad because there are no significant complaints about the direction of the LibreOffice effort. And, I believe, with you, that this fork is wasting resources, and it is dead. It is dead because (all complaints about which license is more desirable aside) there is no likelihood that developers will abandon LibreOffice for Apache OpenOffice.org.

Finally, a word about IBM (and other corporations related to open source development): IBM does not have friends. It has interests. When its interests align with the open source community, we will perceive IBM as “good.” When its interests do not align with the open source community, it will do things we consider “bad.” Remember that IBM, or any corporation, is legally required to do what is in the interest of its shareholders. Govern your relationship with the beast accordingly.


14 Comments

  1. Thanks for posting, although to many of us it seems like stating the obvious.

    The real irony is that due to Apache licensing, code from LibreOffice is not acceptable to AOO. It’s as if Oracle searched to find a way to kill OpenOffice. Akin to them saying,
    “If you won’t work for me for free, I’ll fix it so the name will die, even though I seem like a good guy for donating the code to an open source organization.”

    Apache has been blind-sided. They spent 2 years doing a code review. Alas, there was ill-will created by Oracle and responded to by LOO.

    Many good people stayed with AOO cause they felt it would help bring about what Sun promised in the beginning. Sun never delivered, Oracle would not deliver, Apache can not deliver and IBM is fostering their own interests.

  2. The author affirms his views in a prophetic way. There is not much that supports any of what is written. Like viewing through a crystal ball. Apache has been very successful and nurtures many projects every one knows. Their acceptance of a project is all but done blindly. They just released AOO 3.4-incubating, which works as the original OOO. To say that Apache has been fooled is simply ill driven and nears gossip. Give them more time and let’s judge after 2-3 years.

    • Some in Apache have been fooled. I have read emails and seen it for myself. You call it gossip, but that is just your (incorrect) opinion. For example, I’m amazed at how many people actually think the Apache license is necessary for success in allowing commercial companies to work on this codebase. However, there are many products with a copyleft license that have plenty of commercial success. There might be a tradeoff, but it is wrong to think the Apache license is “necessary.” The existence of LibreOffice is itself proof. Also, a lot of people think that AOO is somehow the natural follow-on of the Sun / Oracle work.

      You don’t consider any opportunity cost in waiting 2-3 years. You also don’t consider that the OpenOffice brand could be damaged. Etc.

      • Please support your views.
        Businesses never operate over a tradeoff. The Apache license is what values most AOO for businesses. There should be no confusion about that.

        The Document Foundation is quite young and we can only judge its success in a few years to manage LO fork.

        The Apache Foundation can be trusted now based on their success.

        • There are many copyleft codebases that have service and support business around it. This idea that the Apache license is necessary or even important is untrue.

          The Apache Foundation cannot be trusted with regards to AOO. You’ll notice that some of the incubation projects have failed. I wrote this: The Apache foundation has a lot of experience, but none with this technology. Therefore, their help will be limited. It is like asking a hospital to fix your car.

  3. This is a bunch of drivel!

    I have yet to see anything that recommends LOO over AOO in openSuSE 11.4. In fact, LOO is so bad I removed it and installed AOO when they released 3.4 on both Linux and windows. Everybody keeps talking about how LOO is so good but I haven’t seen any evidence. As for the font size being in the middle that’s just silly, totally insignificant nonsense and adds nothing to usability.

  4. I’ll wait and see. I have more confidence in pragmatism than ideology and I expect Apache OpenOffice to prosper.

    • What did you think of the HD-DVD / Blu-Ray situation? Of course you’ll wait and see, you have no say in the matter.

      If Apache OpenOffice fails, you will definitely have a learning experience. People need both pragmatism and ideology. And what about the pragmatism around working together? You don’t find it problematic that LibreOffice (i.e. most of the interested third-parties) are against this?

  5. As for whether AOO or LO will win, I was not my intention of predicting it. It merely seems to me that LO is making excellent progress, and I think there is no reason for AOO even to exist. I do predict, however, that, in the end, one of the two will die and the other will thrive, unless they become very different products with nearly-disjoint communities.

    I must also say that I am amazed at the amount of vitriol many Apache-License advocates have for GPL and LGPL. And people say that “free software” people show too much religious-like fervor! If you like the Apache license, use it on your code. If you like GPL, use it on your code. But you don’t have the right to complain that code that you have not written is not available under the license you prefer!
    So, if the license is critical to you, choose AOO or LO accordingly, and hope that its codebase attracts the larger and more active community of developers.

  6. http://people.gnome.org/~michael/blog/2012-04-26-ooo-comparison.html lists the differences between AOO and LibO
    Although I could add some more differences and despite the undeniable fact that some of the listed features give considerable improvements over the common code base, many of the listed features
    – suffer from very sloppy implementation
    – include draw backs compared to the old specs
    – add nothing but MS Office mimicry without really working as in MS Office
    – are not accessible through the API
    – tries to write (and therefore distribute!) OOXML although OOXML describes a vastly different document model than ODF.

    For the majority of every day users both applications are still identical.
    The LibreOffice community has too many fan boys but not enough experienced users. This really huge application suffers from a lack of qualified testers and that anybody can patch anything as long as the result is good looking.

    • I agree that LibreOffice doesn’t have enough people, experienced and otherwise. This codebase, in whatever variant, needs 100s of people. That is why the opportunity cost of this Apache OO effort is so large. The Linux kernel isn’t lacking for resources because of the BSDs and such, but this community is too small to afford to waste people.

      LibreOffice is a new project. There are commonly issues in new groups because of a lack of resources. I disagree that the implementations are sloppy. Features in complicated codebases often have side effects which cause bugs. I think supporting OOXML to some basic degree is a good optional interop feature. Many of their new features look useful to some people, and interesting.

      It isn’t just about the features, but the fact that this team is capable of delivering them. I read the Apache OO mailing lists and see they are perhaps moving 20x slower. I agree that their API / scripting model needs a lot of work. I believe they should adopt Python as the main scripting language and even integrate it deeply into the spreadsheet. Working on such big things will take years, depending on how things go.

  7. Apache OpenOffice 3.4.1 will likely be released this week with many bug fixes (not yet features) from IBM’s Symphony codebase and new language support. Apache OpenOffice is centering around stability while LibreOffice is pointing to new features. Both projects are doing independent cleanups at ortogonal levels.

    Apache OpenOffice is obviously under the Apache License, which has advantages like letting them appear in the Apple store anytime, and partly as a consequence LibreOffice will be abandoning the GPL in favor of MPL.

    This approach (both independent development efforts) may not be ideal but at this point I think both suites will survive and continue diverging according to the interests in their affiliated communities.

    Does it matter? … not really, developers are free to spend their time how they like (which makes me wonder why KeithCu isn’t coding for LibreOffice at all).

    • Of course you are fixing bugs. That is the most basic of tasks. You’ve got a team and they do something. It appears most of them are being fixed by IBM! https://issues.apache.org/ooo/buglist.cgi?quicksearch=From+Symphony

      LibreOffice also fixes bugs in addition to adding features. In fact, many of their features are entered in the bug list. It isn’t a question of whether you are able to fix any bugs, but how many? If you are moving at 10% the speed of LibreOffice, what does that mean? I think it means your project is doomed and you are wasting your time.

      I disagree that the Apache license is better for this codebase. I suggest you read Bradley Kuhn’s writings on the matter:
      http://ebb.org/bkuhn/blog/2011/06/01/open-office.html

      The only reason both will survive is if there are enough uniformed people like you.

      I’m not working on LibreOffice for several reasons:
      1. I’m making a movie.
      2. I’m revising my book.
      3. I worked in Office for 5 years and I don’t want to spend any more time in this problem space in my life. I could talk to you about the special German hyphenation rules. I’d rather just forget all I learned!
      4. I will only program in GC languages. Life is too short.
      5. I see my job as a journalist, I’d rather be a cheerleader than a programmer as I feel that is a more leveraged used of my experiences.
      6. I have some ideas around driverless cars I’d prefer to work on.

      LibreOffice is important to me because I use it to make my book. I don’t need to be a programmer to care a lot about a project.

Leave a comment

Your email address will not be published. Required fields are marked *