It isn't just LibreOffice developers or free software advocates who have queried the motives of Oracle and IBM, or have sensed a desire to play off one community against another (although Apache likes to describe itself as a 'process' rather than a 'community'). Steve Loughran, a long term Apache contributor, noted: "OK, but what about the TDF (The Document Foundation) people. Where do they appear here? Is this another donation of a dead brand after the OSS people forked off the project and kept it going? Like the donation of Hudson to Eclipse after the Jenkins fork?"
"I don't trust Oracle since the Harmony mess", he wrote. "If they did support the Apache license, why don't they provide the TCK for the JDK without imposing Field of Use restrictions? That would be a sign of support for Apache-licensed code?"
The biggest asset contributed by Oracle to the Apache Software Foundation is the brand. OpenOffice.org is a known brand, and most users of OpenOffice.org have no awareness of licensing issues and care even less. The brand will still sell, but the most important asset of OpenOffice.org is not the code or the brand but the community around the code, and the community has gone with LibreOffice.
OpenOffice.org development has been more or less dormant for months and not all of the code donated by Oracle is compliant with Apache's licensing framework. Apache OpenOffice is essentially a new project with a new set of contributors, and it will be many months before it gets into its stride.
IBM has the muscle and marketing power to make a difference, but has shown remarkable tone deafness to the community's reactions, and may have miscalculated.
The objection isn't to Apache or the Apache licence, but to the pitting of one community against another in the cause of IBM's right to re-use the code. As Jeremy Allison put it: "The reason I would like Apache OO developers (including those from IBM) to throw in their lot with LibreOffice is that otherwise you end up re-running the same experiment of Linux vs. FreeBSD. Unless you consider Apple a FreeBSD success (not sure I do, at least from the FreeBSD point of view) then that experiment didn't go well for FreeBSD. But maybe you want to be Apple, in which case good luck (but remember in the best tradition of 'Highlander' – 'There can be only one')."
As this suggests, it isn't entirely clear why the licensing or governance of LibreOffice/OpenOffice is so important to IBM. Lotus Symphony appears to be a fix rather than a long-term component of IBM policy. OpenOffice.org is useful as a placeholder, as a source of components for IBM web office, system management and data warehousing tools, and as a vehicle for the principle of ODF as a universal data format. But IBM isn't in the business of marketing Apache OpenOffice as a rival to its next generation of cloud-based office software, and has no interest in a standalone open source office suite which may disrupt its future market for office software.
One supposition is that IBM sees ODF support and editability, through the use of OpenOffice components, as an important element in its strategy, and can't allow OpenOffice in its current form to die before it has replacements ready to roll out – in which case IBM's interest in the Apache version of the code may be short-lived. OpenOffice is a useful adjunct to IBM's policy of using ODF to replace the control point Microsoft has been able to impose on the market through its proprietary data formats. A reference implementation of the ODF Toolkit written to the Apache Licence is more than useful to IBM, and OpenOffice may be seen as the route by which the toolkit becomes an Apache project.
A secondary benefit for IBM is that Apache OpenOffice is a spoiler and diminishes the market potential for LibreOffice. IBM is happy to push Linux in the server room, but has no commercial interest in the commodification of the desktop.
The interoperability chain
Nevertheless, positives can be found in the Apache OpenOffice/LibreOffice split. Apache OpenOffice may have a future role as a reference implementation of ODF, even if the LibreOffice and Apache OpenOffice forks diverge significantly in other respects. A portable reference implementation of ODF that is released under an Apache Licence can be re-implemented in proprietary and/or free software applications, and gives some assurance of the integrity of the ODF specification. IBM, it should be said, dominates the ODF committees, and Rob Weir, who leads the ODF Toolkit project, has proposed that the ODF Toolkit be adopted as an Apache project.
This may be IBM's key objective and would be a positive contribution, not just for Lotus Symphony or IBM and its greater strategic objectives, but for any competing office suite, and it would allow LibreOffice to focus on the practicalities of upgrading the suite itself, ditching the plug-in philosophy inherited from Sun's CLA problems and other impediments to a smooth working office suite.
LibreOffice has the advantage that it can pick up and include any Apache OpenOffice code, because the Apache licence allows it, although it may be to everybody's advantage if the two projects diverge and offer genuine alternatives – if that happened it would be interesting to see what differences would appear, and why.
Indeed, LibreOffice may have some commercial advantages because it is European and not US based – especially now that SUSE has, at least theoretically, regained its independence, has a close association with LibreOffice and has inherited the community of developers previously employed by Novell.
Whether or not one or other of the two office suites rises above the other, we now have two versions of the same code and two communities working at odds with one another. There is a lesson in this for all concerned – that communities work better and are more effective if, from the beginning, there is little or no ambivalence around the licensing and governance of the code.
For other feature articles by Richard Hillesley, please see the archive.