Apache OpenOffice: who knows where the time goes?

In March Rob Weir put up a blog entry entitled "Where did the time go? (A look at the Apache OpenOffice timeline)", showing the slow progress of Apache OpenOffice through the process of acceptance into the Apache fold. Richard Hillesley looks at the issues…

Oracle announced the donation of the code to the Apache Software Foundation in June last year. For a number of reasons, primarily Apache incubation issues, code migration and removal of copyleft, there has been no update of the OpenOffice codebase since.

Last month IBM announced the release of Symphony 3.0.1, which will “be the last release of IBM’s own fork of the OpenOffice codebase,” and has declared that IBM’s future energies are to be devoted to Apache OpenOffice, of which IBM expects to release an exclusively IBM edition, which promises to be “an identical release of the Apache OpenOffice code under the Apache license,” with plug-in extensions (primarily to plug in to other IBM products).

Significantly, the pledge to release the code of Lotus Symphony to the Apache Software Foundation, made last July, is yet to be realised, perhaps because of a combination of ongoing code development and relicensing issues.

There has been a lot of speculation about IBM’s role in Oracle’s donation of the OpenOffice.code to the Apache Software Foundation, and its decision to follow the code to Apache, rather than work with the community on LibreOffice. It is usually suggested that the decision was driven by licensing issues. A liberal licensing regime, such as the Apache licence affords, allows the office suite to become a repository of useful components for web office, system management and data warehousing tools that reach far beyond the scope of a stand alone office suite, and there is no obligation to give changes back to the community.

The unhappy result is that there are now two versions of the most successful open source office suite, a community fork, developed primarily by SUSE and Red Hat engineers, and an Apache fork, which is effectively run by IBM engineers.

LibreOffice is thriving, and has many more committers and commits than the Apache project. Most of the contributors to the Apache project are IBM employees, and LibreOffice appears to have attracted more of the original Oracle/Sun employees. The core LibreOffice contributors who are not employed by SUSE and Red Hat are members of the wider community.

At the same time removal of code that is licensed exclusively under copyleft from Apache OpenOffice has necessitated the removal and/or replacement of many features, such as filters (libwd), complex-text rendering via ‘Graphite2’, removal of webdav support, the calc solver, PDF import, wordperfect import/graphics, and the database report builder; desktop icon integration is degraded and spell-checking has had to be replaced due to the loss of hunspell.

A curious side effect of this process is that community developers who licensed their code under the LGPL but assigned their copyright to Sun Microsystems, now find their code being re-licensed, against their will, to the Apache licence, which may mean the later absorption of the software in proprietary offerings.

Conversely, IBM personnel have hinted at ‘patent issues’ in the LibreOffice code:  “One thing not to lose track of is that Symphony has done IP remediation at many levels… But I know with certainty that we’ve fixed things that LO has missed. (I’m talking patents, not the MPL/LGPL dependency issues).” There is no word on what these patents might be, or indeed whether the issues are real.

While LibreOffice makes steady progress, it is hard to anticipate when a release of Apache OpenOffice will be available for people to play with, to compare and contrast with the community project.

It has even been suggested that OpenOffice may fail the incubation process which inducts the code into the Apache Software Foundation. Not all Apache contributors were happy with Oracle’s donation of the code to the ASF and the snubbing of the community that had worked on the code for the previous ten years. The abandonment of the community echoed Oracle’s donation of the Hudson project to Eclipse and the abandonment of the Jenkins fork of Hudson – “From the whispers of ApacheCon, may never leave the incubator project. The intention may be to do a thorough code audit and produce one last, clean release that the rival LibreOffice can absorb.”

The split is wasteful and divisive and favours nobody. It would be no disaster if the Apache OpenOffice project were to fail to incubate, and IBM were to join Intel in giving support to LibreOffice, but nobody is anticpating this any time soon.

For the time being LibreOffice has the wind behind it, and the IBM project is trailing in its wake.