Editor's Column

9 January 2001

The 613 build:  problems and opportunities

On 13 December 2000, Sun’s engineers released the 613 installation set (the latest build) to the OpenOffice.org community.  As with all of previous builds, Sun’s engineer’s had thoroughly tested the software on the standard assortment of boxes and under the usual circumstances.  The release performed as designed, and there was no reason to suspect that any problem would be worse than the usual for a pre-alpha release.

For these reasons, then, the bugs that were immediately discovered by community members who attempted to install and run the software were all the more surprising. The bugs rendered the software by and large inoperable and unusable. These community members were not, it should be stressed, using strange boxes; their setups were perfectly ordinary. What is more, all previous builds installed as expected and were stable.

The traffic in the "discuss" mailing list was far greater than average, and members were busy offering warnings and advice.  But it seemed that many community members could not successfully install the build, let alone run the software.

Once the extent of the problem became clear, pulling the installation set became a plausible response. From this perspective, the 613 build reflected an inexplicable and slightly embarrassing slip-up in quality assurance that should not be inflicted on the community at large. Indeed, barely twenty-four hours after the release, Sander Vesik initiated a new thread in the "discuss" list that asked the community members whether the installation sets should be pulled. At the heart of the query lay the concern that community members were both wasting their time with 50-plus megabyte downloads and that the OpenOffice.org project was revealing itself to be an unprofessional effort.

The response to Sander’s poll was immediate and at first seemed to lean toward the idea of pulling the installation set, with one correspondent arguing that removing it would "save face." But then OpenOffice.org Community Manager Adam "Goolie" Gould persuasively argued that saving face was not the point of an open-source project: 

"If this were a proprietary beta release," Goolie pointed out, "then yes, we all wouldn't want to 'look bad' in front of our customers, and would be worried about damaging our reputation in the market."  But the build was not a proprietary beta release, and OpenOffice.org is not a proprietary project. In fact, OpenOffice.org is an important leader in the open-source movement and, rather than illustrating a failure of QA, the problems surrounding the release and the community response indicate the strength of the open source model for finding and addressing development problems.

That’s because, "As far as Sun could tell, they released a perfectly fine build; it worked on all their test cases.  We only discovered problems when you, the larger OpenOffice.org community, discovered some issues on your unique "test cases".  This is exactly when we need your help." The great advantage, in short, of an open-source community lies precisely in its ability to spot (and even fix) bugs of this sort.  Pulling the build would thus defeat the purpose of OpenOffice.org, which does not pretend to release bug-free builds and relies on the dialectic relation with the community members to create better software.

The installation set was not pulled. But, prospective downloaders were warned that the build was problematic.  Has the project suffered as a result of the problematic build? There’s no easy way to tell. But, judging from the level of developer engagement, measured by mailing-list posts and downloads, the answer is, no.  In fact, for the week of 17 December to 23 December, there were a very impressive 4,310 downloads. The numbers suggest, in short, a community hardly scared by the problems of the build and instead even more engaged, even more interested in the future of OpenOffice.org. 

My next column addresses a point recently raised in the discuss mailing list and that can be summarized as, Whither OpenOffice.org? or, Where do we go from here? And, as always, if you have any questions, suggestions, or comments, please, don't hesitate to e-mail me at the address below.

--Louis Suarez-Potts (louis at collab.net)*

* By spelling out the mail address, I'm hoping to defeat spam crawlers.