Rob Weir explains the OpenOffice resurrection | Interview

Last week, we had the first OpenOffice release, since the project was donated to the Apache foundation. This raised a lot of questions and many users wondered what is the point, or what is the difference with the Libre “brother”? In an attempt to answer these questions and learn more about how the people of the Apache foundation resurrected OpenOffice, we meet Rob Weir on this Monday interview. Enjoy!

Introduce yourself and tell us a few things about what you do for OpenOffice.

Hi, I’m Rob Weir, a Senior Technical Staff Member at IBM and a Committer on the Apache OpenOffice and Apache ODF Toolkit projects. I joined IBM as part of the acquisition of Lotus Development Corporation back in 1995, where I was working on SmartSuite. (Remember 1-2-3?) So I’ve been working on office applications and related topics for most of my career. I am also Chair of the OASIS Open Document Format Technical Committee, the standards committee that controls the ODF document format standard.

At Apache, volunteers are able to focus deeply on a single topic of interest if they wish, or work more broadly on a large range of project concerns, or anything in between. We don’t have fences that box someone in (or out). That’s something I really like about Apache, since I like to dabble in everything

Most recently, as our Apache OpenOffice 3.4 release went out, I’ve been working on writing up bulletins for security vulnerabilities fixed in that release, writing scripts to report on 3.4 download numbers and thinking about how we can better scale our releases to handle more languages and platforms.

How and why exactly did the Apache Foundation end up with OpenOffice in its hands?

How it got to Apache is easy: Oracle signed a Software Grant Agreement (SGA) that put the code under the Apache License 2.0. The Apache Incubator project then voted to approve OpenOffice as a new “podling,” which you can think of as sort of a probationary project, where communities that are new to Apache (even though they may be quite old in their own right) are given some extra help from more experience Apache members (“Mentors”) as they acclimate themselves to Apache tools, procedures and The Apache Way. A podling is also where the project gets their code into conformance with the Apache License. At the same time an initial set of nearly 100 volunteers were invited to become members of the “Podling Project Management Committee,” the body that provides oversight for the direction and operation of the project.

Oracle also transferred the domain names and registered trademarks to Apache. They then actively worked with the Apache OpenOffice project to transition the various Oracle-hosted services over to Apache, e.g., community forums, issue trackers, website, wiki, etc.

“Why exactly did this happen?” – This was the result of decisions and actions by many players, including some major corporations, but also a non-profit foundation and hundreds of volunteers. Motivations no doubt are as diverse as the participants are. But this much is certain: a large number of people wanted this to succeed and have backed it up with their actions and efforts.

In any case I will say that I was delighted that Oracle did contribute to Apache. When IBM acquired Lotus, I saw SmartSuite disappear, lost to the world. The efforts of 100’s of developers, a decade of work, was locked away in some corporate archive. That’s the way it was done back then. But with open source we now have the opportunity for a code base to outlive the interests of any one corporation. We’re not tied to a single company’s business model and profit/loss statement. We’re insulated from the gyrations of markets, mergers & acquisitions, etc. We’re now at the Apache Software Foundation, one of the most (if not the most) mature and respected open source foundation in the industry. This is a great place to land.

How hard was it for the Apache members to decide and accept the challenge to continue the development of OpenOffice?

Apache does not operate in a top-down hierarchical way, determining the technical agenda of its individual projects. That’s not how it works. The Foundation provides a governance framework, infrastructure, a license and a set of cultural conventions for community-based decision making we call “The Apache Way”. So the ASF did not so much decide to “continue the development of OpenOffice” as decided to accept a proposal for a sizable number of volunteers (nearly 100 at the start) who wished to create a “podling” in the Apache Incubator.

The project is a non-hierarchical, flat organizational structure. So there are no formal teams and therefore no formal titled leaders. Leadership is more about “taking the lead” on something you want to do, and enlisting others to help with that task. I might take the lead on security today, and someone else tomorrow. That said, there are some recurring topics that are persistent enough to have spawned up their own mailing lists within the project – the closest we have to subteams. These exist for QA, Marketing, Security and a few native languages, e.g., German, Italian, Japanese, etc.

What was the impact that the Oracle era had on the community around the project?

Oracle has been active, both in the project and behind the scenes in transferring over source code, domain names, trademarks as well as transitioning the online services to Apache. This effort – thanks to Oracle’s Andrew Rist and the Apache Infrastructure Team – has been very successful.

As for the community, almost all of our volunteers are former volunteers, many of them with a decade or more experience with that project. And now that Apache OpenOffice 3.4 has been released, we’re getting even more seasoned volunteers coming back, especially translators.

We had the OpenOffice 3.4 release recently, the first that came out of Apache foundation. This raised a lot of questions from many people. The big one was, why choose OpenOffice instead of LibreOffice?

Why not Microsoft Office? Why not Calligra Suite? Why not Abiword or Gnumeric? Why not iWork or WordPerfect?

When you think of it, there are probably one billion people on this planet who have a use for an office suite. Given the wide choice we have, in free market economies, of different brands of automobiles to toothpastes, it would be odd if there was less choice in office suites. Perhaps our experience with Microsoft Office in the past decade has made us satisfied with monopolies? I hope not.

The fact is, user preferences are not all the same. Some are “over-served” by Microsoft Office. They are paying for more features than they need. They need only 30% of the features, but there is no commercial offering that meets their need. Others are “under-served”. They want more features than MS Office offers, but they cannot buy these features at any product, so they need to pay for custom development. Having an “one size fits all” approach only serves the median user, but fails to account for the diversity of needs that exist in any real marketplace.

Food for thought: I’d recommend this TED Talk from Malcolm Gladwell, discussing consumer preferences and satisfaction in the context of spaghetti sauces;

Much of the same analysis applies to software. We need to get away from the belief that monolithic projects with hoards of developers is the best or only way to satisfy more users.

Is there any chance we will see some kind of collaboration between you and LibreOffice developers? In what sectors would something like that be beneficial for both of you?

We already have a strong working relationship and ongoing collaboration with LibreOffice in the area of security and vulnerability analysis and reporting. Of course you do not see that, since such work is by necessity done on private lists.

There is also ongoing and active collaboration, along with other commercial and open source apps, on the further development of the Open Document Format (ODF) standard.

Apache hosts a community support forum, which supports OpenOffice users as well as LibreOffice users, and is staffed with volunteers from both projects.

And we have developers who happily contribute patches to both projects.

So there is some collaboration already. But there is certainly room for improvement.

Could the two big Linux office suites collaborate, with the one taking the role of the “serious” and stable, while the other could work on new things completely free and riskless?

My guess is that both projects will end up with two release streams, an LTS one for those interested in stability, and a trunk release for the “latest and greatest” code, but perhaps less stable.

Right now, what are the biggest differences between LibreOffice and OpenOffice? Also, would you say that the two development teams lead the projects technically away from each other?

The projects and the products, of course, are closely related, siblings in a sense. So the resemblances are many and the differences are few. The main differences are differences in license and a handful of minor feature enhancements that either project can claim as unique in their version.

It is natural that the code bases will diverge as time goes on, based on the non-identical interests of each project’s volunteers. From what I can tell, looking at expressed interests of those in the Apache project, we’re likely to see more platform support, more UI (and accessibility) enhancements, and a greater emphasis on stability and enterprise requirements.

Did, or do you have any kind of contribution from employees of Oracle that worked on OpenOffice before?

Absolutely. In fact several of our core developers are former Sun/Oracle employees. We also have programmers from the Lotus Symphony team, who are experts in the code from their work on the Symphony fork. And recently we had several engineers from CS2C in Beijing join the project, with experience in this code base from their RedOffice fork, especially with CJK layout issues and UOF file format support. So I think we have more experienced developers with this code base than any other project can claim.

Do you test the application with both Oracle Java and OpenJDK? Is OpenOffice working equally good with both?

Yes. AOO uses the system JVM. We’ve tested both Sun/Oracle and OpenJDK. I’m not aware of any issues.

With OpenOffice being a mature project, I understand that it is natural for you not to take big risks and play the development safe. Are you planning any big changes though, towards the evolution of the project especially in the user interface sector?

Standing still is risky as well. On the UI side the game changer here is the contribution by IBM of the Symphony source code. The UI of Symphony has been widely praised, for example winning the Editor’s Choice award from PC Magazine for the last two years, beating out both and LibreOffice. So I see a huge opportunity here.

Are you planning a cloud release for OpenOffice, similar to Google Docs?

Something like that would probably not occur in the Apache OpenOffice project. Apache discourages “umbrella” projects. IMHO, it is better to have a cohesive, non-hierarchical community working on a single code base. If someone wanted to develop a cloud office at Apache, it could certainly be done, but I think it would make more sense to do it as its own Apache project.

What are the plans from now on? How often are we going to see new releases and what features are in the top of the list right now?

Right now we’re working on a near-term 3.4.1 maintenance release that will fix some high priority bugs that were identified in 3.4.0, as well as add support for additional languages, likely UK English, Hebrew, Finnish and Norwegian and maybe others.

We’re also discussing the next major release. Whether this is a 3.5 or a 4.0, and what features are included – this is all up for discussion on our project list. Ideas that I think are interesting include adding support for standards like CMIS, OData, OpenSocial and UOF 2.0, and integration of Calc and R statistics.

Also, recently volunteers with professional usability experience have come together to study OpenOffice usability, so I expect to see good things coming in the future.

There is also active 3rd party porting work for *BSD, Solaris and OS/2, some of it being beta tested now.

Thanks Rob! It was great, getting to know more about one of the most useful and most used applications in the Linux world.