Off the Beat: Bruce Byfield's Blog

The rumors about LibreOffice that I mentioned a couple of weeks ago apparently lack substance.  Andrea Pescetti, Apache OpenOffice PMC Chair, states that, "I can say that I don't witness the reduction of activity you describe: community members (including those that I know to be employed by IBM) are all participating in conversations and development." So, I am left wondering: how can the relative health of Apache OpenOffice and LibreOffice be measurde, four years after the two parted ways?

OpenOffice representatives are fond of citing download statistics -- possibly because theirs are available on public sites, and LibreOffice's are private. However, either way, downloads are a poor metric, because the same person may download multiple times for different machines, operating systems, or releases, depending on their habits. For instance, since OpenOffice.org's code was first freed in July 2000, I have easily made one hundred downloads by myself -- maybe more.

Anyway, the software's popularity is at best an indirect measure of a free software project's health. A better indication probably lies in measurements like the number of people involved, and the changes to the code.

Mailing list indicators
One measure of relative project health is the number of messages on their mailing lists. Although both LibreOffice and OpenOffice undoubtedly work through some form of chat, like OpenOffice.org before them, both emphasize mailing lists for communication.

 As with the number of downloads, the user mailing lists are of limited use, because they measure exterior interest rather than internal activity. However, those users who find their way to a project's mailing may be more likely to be involved with the projects than the average downloading user. And, in fact, many regular project contributors do sometimes post on the user lists. So, it may indicate something that LibreOffice's user list for the past month averages 20-25 messages per month, while OpenOffice's average 5-10.
 

 Almost certainly,  development lists are probably more accurate indicators. Comparison is not always possible, because the two projects do not always use the same list names. Also, LibreOffice divides development into more specialized mailing lists than OpenOffice does, which could mean that its members are under-represented on the general lists.
 

 However, the general development lists are roughly equally in the last month, while in September, LibreOffice had about twice as many messages per day as OpenOffice. Similarly, in the localization lists, both averaged about four message per day in the last month, although OpenOffice dipped below one per day in the previous month. On the documentation lists, LibreOffice averaged 4-6 messages per day, compared to OpenOffice's two messages in the entire month.
 

These figures are typical for most of the past year. They seem to suggest that, just as its supporters claim, LibreOffice is the most active of the two projects. The figures may also indicate the projects' priorities.
 

The Open Hub stats
 
 Statistics from Open Hub (formerly Ohloh) support the contention that LibreOffice is the more active project. In fact, they suggest that the evidence from mailing lists is extremely conservative.
 

 According to Open Hub, LibreOffice has 298 current contributors to its code. From the fact that the news release for LibreOffice 4.2.5 claims "over 800 contributors," I infer that the project has over 500 non-coding contributors.
 

 By contrast, Open Hub attributes 39 current contributors to OpenOffice's code. I could find no figures for OpenOffice's non-coding contributors, but Open Hub lists 124 contributors total. Unfortunately, the total number of LibreOffice contributors is not comparable, because LibreOffice includes OpenOffice.org contributors while OpenOffice does not -- yet, even so, the difference is startling.
 

 Not only does OpenOffice appear to have lost two-thirds of its coding contributors, but its current contributors are outnumbered nearly eight times by LibreOffice's. Since OpenOffice does not have any localization, art work, or usability accomplishments that exceed LibreOffice's, the situation can be presumed to be no better among non-coders.
 

 Yet for me, the most telling difference is in the lines of code. Both projects grew from OpenOffice.org (and zealots can be found to claim that both are the so-called true successor). But, despite this common source, LibreOffice is listed as having 7.2 million lines of code, while OpenOffice is listed as having 11.2 million lines. In other words, LibreOffice has not only managed to release a steady series of new releases, each with incremental changes in the features and the interface, but also to remove over 35% of the code while doing so. More than any other, the implications of this single statistic shows the relative health of LibreOffice and OpenOffice.
 

Cumulative problems
 
 The more you look at the two projects, the more the fact that LibreOffice is outperforming OpenOffice becomes undeniable. The difference has nothing to do with the quality or development of the code; after three years, the differences are only now starting to become visible to average users.

However, OpenOffice simply has too much against it. It lost months of development time while Apache reorganized the project and did a code audit, and the problem that licensing differences allow LibreOffice to borrow its code while it cannot borrow LibreOffice's has no obvious solution.

Yet the problem of participation appears even more serious. As imperfect as the indicators I use undoubtedly are, they show no area of development in which OpenOffice outclasses LibreOffice. Even more seriously, OpenOffice is often far less active than LibreOffice -- in particularly in attracting developers.

None of this discussion has anything to do with the quality of the code. OpenOffice's fundamentals are identical to LibreOffice's. Yet with its cumulative handicaps, OpenOffice seems destined to slowly lag more and more behind unless it can rally for a massive effort to find more contributors.

Without such an effort, OpenOffice will continue to struggle on. Free software being organized the way it is, there is no reason for the project to disappear. Nor is a major reversal impossible. Just now, though, OpenOffice is increasingly resembling a wounded animal, bravely staggering onwards until it finds a lonely place to die.