Alex Hudson

Thoughts on Technology, Product, & Strategy ⇢ Apache

Many words have been expended on this situation. I don’t have an awful lot to add about the project side of things: I think it’s immensely sad that is being forked again (this is much more clearly a fork than LibreOffice was), but fundamentally all actors within the free software world are autonomous and have free will. Such is life.

(this is a deeply opinionated blog post. feel free to skip it, take it with a grain of salt, whatever.)

A lot of people are talking about this as mainly a license issue, with one commentator spelling it out: “It will be interesting to see, after the first wild set of commentary flies, which project – and which license – that various developers and corporations alike choose to actively support” (the juxtaposition of the opinion and the title of the blog is especially delicious). After all, it’s easy to see this as a license show-down: given the same code base, and two different licensing strategies, which will succeed? Sadly, it’s not really a race LibreOffice can “win”: if that project is more successful, well, they had a head-start, didn’t they.

I don’t think much of this is really relevant. I’m not sure IBM care too much about who develops the code, and I don’t think an LGPL’d code base would fundamentally stop them from shipping a proprietary product if that is what they wanted to do (it makes it harder, of course). I actually think this is all about OpenDocument Format, which is a subject virtually no-one has raised.

If you look at the OASIS TC, you can see it’s pretty obviously dominated by Oracle (was: Sun) and IBM. There are a few representatives of various other companies and open-source projects, but fundamentally this is a closed shop with a pay-to-play rule which means you have to pony up to join. The v1.2 spec – which has been used by since 2008 – has only just managed to crawl out as a committee specification, incredibly late. What this means for OpenDocument v1.2 documents as read/written by 3.0, who knows. But with Oracle fading into the sunset (sorry), large chunks like OpenFormula finally done, maybe v1.3 will actually show up on time.

So here’s where this really matters. ODF is nothing if not a stick with which to beat that other vendor. And having a large say in ODF is not really anything unless you have implementations around, and until now, has not only been a leading ODF implementation, but effectively the reference implementation of ODF. This is what IBM really want control of: this is why they’re so involved in the TC, and this is the reason for the sudden outburst of passion for (all in my opinion, obviously). They almost cannot allow LibreOffice to obtain the de facto mantle of “reference implementation”: ODF as a standard is supposed to follow implementation, and obviously IBM (and most of the rest of the TC, for that matter) have no skin in the LO game. Worse, I don’t think they want LO adding and changing features  in ways which necessarily touch ODF: imagine if it had been LO only who had released ODF1.2 support in 2008. This is a very real risk – there are still critical features missing from ODF (want CMYK support do you?!) and control of the TC alone means precious little without the reference implementation.

Is this going to be a bad thing overall? My suspicion is that this isn’t going to be a terribly short-lived fork; I think IBM’s decision to move to Apache (well, Oracle’s, but we know whose advice they were following) is going to have a long term impact. It’s already pretty easy to see how the divisiveness has taken hold:

  • the AOOo project has recently discovered that used to solicit donations, and don’t sound amazingly thrilled about how it was set up. It’s quite intriguing to watch people like Rob Weir claim that the community has moved to Apache, when the people at Apache seem to have little idea about how OOo operated, where money came from, who spent it, etc. It’s a sad day when bystanders to your project (such as myself) appear to know more than the project leaders. (If someone on the Apache list could also let them know that the SPI holds c. $20k on their behalf, that might be a good idea – they don’t seem to have figured that at the time of writing). Anyway, the likelihood of you bring able to donate to one non-profit which supports both projects (as you can right now) seems to be diminishing fast.
  • there’s been a lot of talk about AOOo code somehow making its way into LO. I strongly suspect this isn’t going to happen that much either: I think these projects are going to diverge must faster than people realise. LO has already moved ahead and dumped a lot of old code; by the time AOOo actually gets off the ground this will be even more so. When AOOo arrives, what core features is it going to get early on? I imagine IBM will make various “donations” – they’ve already talked about accessibility, pivot tables, file format stuff. Some of that may get pulled in, other bits (like pivot tables) is probably going to be too different. After that… I think there will be some work to integrate AOOo with other Apache projects, like POI, and Apache people are quite happy using Java. I don’t think LibreOffice will want much/any of that. It’s going to take non-IBM AOOo hackers time to get up to speed enough with the core to make interesting changes, LO will be years down the road at that point.

There’s enough momentum behind AOOo that means it’s not going away any time soon. I rather suspect there isn’t enough momentum to reach “IBM escape velocity”, and sans some rather large contributions that LO is unwilling to take (which is quite possible – even for potentially desirable stuff like the Symphony UI changes) it’s difficult to see the project making the same strides LO has. More likely, there will be various “quick wins” early in the project but not much else.

It would be interesting to see someone write a full history of StarOffice from the very beginning. It has always been a project full of potential, and one particularly important to the Linux desktop even though its main userbase is on Windows, but it’s also a particularly political project in terms of what it attempts to achieve. As such it has been turned into something of a football.


I voted in the Fedora 2011 elections


Short thoughts on the riots.


  1. Hey, I like delicious juxtapositions, even if I’m not sure that I’m seeing the same juxtaposition that you are.

    Also, if you have the specific contact information for people at SPI and the like, why not send it on to the general@incubator list? Or send it to me and I can send it to the list. It’s no surprise that the ASF’ers working in the Incubator don’t know about all the details since the Incubator only heard about this donation last week; I think this is something that took everyone by surprise.

    And re: “IBM escape velocity” (great quote), I count 31 OOo-related email addresses on the initial committer list in the current document. Now I’ll admit I do not know the detailed community history around, but that certainly sounds like there’s goodly amount of non-IBM’ers who want to be working on this.

    Thanks for the commentary and the link BTW; it’s nice to see deeply opinionated posts that at least don’t feel like they’re attacking the Apache Way as some others seem to be posting.

  2. Duv

    Now that you mention OASIS, that does beg a bit of a question… does The Document Foundation sit in the OASIS TC in any kind of manner (by membership via it’s partners [Red Hat, Canonical, etc.], by direct membership, or othermeans)?

    From my thinking, to have some insight on ODF’s direction… that would be the place to be. If TDF, isn’t on that Committee… it begs the other question of why aren’t they on the TC or do they have plans to join the TC in the near future?

    Pardon my random rambling.

  3. Alex

    @Shane: well, the comment about not knowing what was going on wasn’t really aimed at the ASFers per se: I thought it was more indicative of the state of the project as it is being brought over, essentially bereft of the culture and knowledge of many aspects. Watching people trying to figure out how much bandwidth is needed, what compile farms could be used (I mean.. Hadoop… really…) – all this sort of stuff should be the easy bits.

    As for the non-IBMers, I think it’s going to be interesting to see what amount of code comes that way. It’s quite easy to get translations and other stuff coming in. Fundamentally, though, OOo is not a codebase amenable to drive-by patches. 30 non-IBMers is not even nearly enough in my opinion, if you imagine what the split in skills is going to be.

    I personally have no problem with the ASF attempting to take this on; I just don’t think Apache really knows what it’s getting itself into. I think the whole thing is just going to be a massive drain on resources – money, goodwill, whatever – and I struggle to see the upside for ASF. I also don’t think it’s a cultural fit at all. But, at the end of the day, that’s what incubation is for, and as long as people +1/-1 objectively, then there’s no real problem.

    @Duv: officially, no, they have no representation. Whether or not they will want it, I don’t know, but certainly there are LibreOffice contributors who are involved there. Whether or not it says anything about the direction of ODF – I really don’t know. The standard is so slow-moving right now the question of whether it will even be relevant in five years’ time is something of an open one.

  4. MizK

    A sad, sad, sad tale for a wonderful product and organization.

    Re history– grew out of StarOffice, nee StarDivision….some chronological details…

    StarOffice milestones

    To have OpenOffice end up as an Apache “incubator” project is an insult beyond words! This is a full-fledged product!

  5. Alex, the membership of the OASIS ODF TC is listed here.

    The names to pay attention to are the “voting members”. Those are the members who are active in the TC’s work. IBM has 3 voting members, which represents something like 15% of the votes on the TC. I see no basis for your assertion that the TC is “pretty obviously dominated”.

    You asked about TDF membership. Although participation in the technical committee is at the individual level, with one-person/one-vote, individuals as well as organizations can join OASIS.

    TDF would be welcome to join, as have other open source projects, like KDE. But it probably makes sense for them to first establish their foundation, so they have something that can join as an organization. Until then I see that two TDF Steering Committee members are also ODF TC members: Charles Schulz and Thorsten Behrens, with Charles joining via his company, Ars Aperta, and Thorsten via Novell. Both have been long standing and valued members of the committee.

  6. @Alex: ah, thanks, that makes more sense. Yes, it is a significantly new kind of project for the ASF, but that’s why we have a lot of volunteers.

    @MizK: The point of the Apache Incubator is to ensure there’s a sufficient *community* that is willing to work in the Apache Way to become a top level project at the ASF. Incubation is a process for the community, not the product itself, which indeed already has a major effect both on technology and users who are running it.

    Incubating takes nothing away from that, and indeed doesn’t take anything away from any other communities – like the TDF – who wish to take all the code, now that it will be under the Apache License. Incubation ensures that the community is healthy and diverse (i.e. not dominated by one employer) and is able to function independently in the Apache Way. At that point the Incubator PMC votes again to graduate the podling into a top level Apache project. And note, the podling will have the chance to choose it’s name during the incubation process, so this may be more like a rebirth of a new project rather than a simple transfer of the existing project. It all depends on who shows up to do the work of the podling.

  7. Alex

    @Rob: well, there are a couple of issues here that you’ve glossed over.

    First, “dominated by” does not imply “also holds a voting majority”. 15% of the votes on the board of any company is a very strong position indeed, and my comment on domination was not limited particularly to voting.

    Second, I’m not sure what OASIS rule you’re referring to when you say the TDF would need to be an incorporated association. Maybe there is such a rule, I don’t really see anything online that says TDF couldn’t join right now as they are.

  8. Alex, You’re misinterpreting the membership roster. IBM has only three voting members because we only have three employees participating. Just three of us attend meetings, participate on the mailing list, etc. The other names you see participated back in 2006 when we were working on accessibility and ODF 1.1. On the TC IBM is vastly outnumbered by individual members. We’re outnumbered by Microsoft + Nokia, if they wanted.

    15% gives IBM no particular advantage. I’m not sure what alternative means of “domination” you might suggest, if it does not involve voting, writing or speaking.

    As OASIS membership, if you read my words as carefully as I’ve written them, you’ll note that I never said a foundation was *required*. I merely said “it would make sense”. I’m thinking about transferability of the organizational membership. TDF should check on that lest they waste the membership fee.

  9. Alex

    @Rob: two things. First, my comment about domination was not about IBM alone, so please don’t attempt to put words into my mouth about claims I haven’t made about IBM. Second, I wasn’t making a statement about voting rights, as I said previously. I wasn’t making a statement about the membership roster either, so it’s difficult to see how I could possibly be misinterpreting it.

    If you’d like to go on some statistic exposition about how the ODF TC has not been dominated by IBM and what is now Oracle feel free, but this observer disagrees. How it may play out in the future, of course, is something totally different. It’s difficult to see members like Oracle, Microsoft and Nokia putting in a great deal of effort to move the standard forward…

Leave a Reply