Safe as Milk

Dealing with system integrators

I was interested to read the exchange around Matt Asay’s “dealing with system integrators” post last week.

What it comes down to is when you’re a free software producer, if you want third parties installing and commercially supporting your software to partner with you, you need to offer something in return.

If my company wants to have some free software commercially supported, I would laugh any integrator who suggested buying a subscription to the commercial version, and who refused to start working on integration until I’d purchased if there were no benefit in that for me. I would insist that the integrator installed & supported the Free version, unless there were very compelling features I needed in the commercial version.

If Alfresco had tied the hands of the guy in front of me behind his back, I would go find someone with no such constraints to support the software.

So for the three people involved in an official integration partner deal (Alfresco, the integrator, and the client), all three parties must have some interest for the deal to happen.

The client wants cheaper, high-quality software. If the commercial offering gives compelling features, he may pay for it. If it doesn’t, he’s not an idiot.

The integrator wants to make some money on the support deal. If being an Alfresco partner gets him more leads, more deals, more clients, more money, then he’ll go for that, and will pay an annual partnership fee, your annual training seminars and any other costs you associate with being a partner. If partnership is a mill-stone around his neck restricting what he can offer his clients to keep them happy, he won’t.

Finally, Alfresco, the software producer, wants as many copies of their software installed worldwide, and they want to ensure high quality local support (so they have an interest in having a high-quality partners network).  They also want to maximise the number of installations of commercial versions, and thus have an incentive to ensure that the commercial offering is compelling for people who will be paying for support.

Forcing integrators to sell commercial copies of the software before doing integration doesn’t in itself make those versions compelling. In fact, if you force extra conditions like the ones Matt outlined, you run the risk of helping a renegade unofficial Alfresco support network to spring up. You’re selling free software, they’re free to support it without going through you. If that happens, you have a short-term gain for a long-term loss.

Getting constructive about Nokia and maemo

There has been some criticism recently of Nokia and its handling of maemo - improving the state of affairs is one of the reasons why Quim contacted me and asked me if I’d be willing to work with the project to improve things.

The key to solving any problem is straightforward:

1. Identify and characterise the problem
2. Address one after the other the root causes of the problem
3. Evaluate the situation after each change

This is similar to Federico Mena Quintero’s characterisation of profiling code. In fact, a surprising array of problems are suitable for attach with measure, change, re-measure, rinse, repeat.

The over-riding arc I’ve been hearing so far is “Nokia is hoarding control over the project, and aren’t doing enough to help the maemo community”. I think that’s a mite unfair, and often I get the feeling that people on the mailing list are confusing a reality where there are problems, but they are poorly characterised, and malicious intent on the part of Nokia.

Some examples:

• No, Nokia isn’t all-powerful, and can’t make Google fix Reader so it works better in microb
• Parts of the platform are restricted and can’t easily be replaced with later versions. Let’s get an explanation for that, and talk to the right person to get it fixed. Right now we don’t have the reasoning behind the decision, and that’s what’s missing in characterising the problem
• “Nokia is keeping control of the project” - what are Nokia keeping control of? Let’s identify the list of resources that would be useful to community members, and work, one by one, on seeing if Nokia is actually keeping control of them
• “The N8×0 tablets ship with proprietary components” - my priority is to ensure that you have documentation for everything possible on the tablet, but to me, there are two different things, maemo, the community project, and the N8×0 tablets, which are commercial ventures using maemo [*]

So I plan to apply this optimisation technique to various problems in maemo. For each proposal I make, I will be looking for feedback from Nokia and the maemo community to see if it is a step in the right direction.

For a start, I will be proposing policies for access to maemo resources, including maemo.org email addresses, the maemo trademark (which is of course linked to the email addresses), and any parts of the maemo platform which community members don’t feel are sufficiently open.

My goal is not to get everything open in doing this. It is to make clear the limits of the maemo project, and in this way ensure that expectations on both sides of the equation are coherent. I hope that Nokia will accept the proposals I make, but even if there are arguments against, I believe those arguments can be open, and clearly understood by all involved.

* I don’t believe that there has ever been ambiguity about this - all of the tablet which can be open is open, but some decisions to use closed components were made in the interests of product differentiation, cost and other reasons. The N8×0 is not meant to be a completely Free product (unlike the Neo1971 or OLPC, which do aim to be completely Free). What we should insist on is that someone buying an N8×0 has all of the tools they need to paperweight it with custom, non-Nokia, software, and access as much of the hardware as possible with free software.

OpenOffice.org - a candidate for a 501(c)6?

Following up from my previous post, there has been some interesting discussion in the comments and elsewhere. One issue in particular came throughh in a couple of comments.

Ted Ts’o is gushing in his praise of the Eclipse project:

Look at Eclipse; it was released by IBM in November, 2001. Within 2 years, it had something like 80 companies participating in the code development, and in less than 2.5 years, a non-profit organization was founded where IBM didn’t even have a majority of seats on the board.

And Michael Meeks  brings up the subject of OpenOffice.org governance (which he has written about frequently in the past - by the way, Michael, I can’t find an easy way to link to individual journal entries of yours):

Faced with serious, persistant maladministration and injustice in the ‘communities’ Sun controls - what can you do?

Indeed, Sun has a ways to go make sure that projects they free have a non-negligible contribution from people outside their organisation, and OpenOffice.org is probably the most compelling case for an independent non-profit that they have right now. You have all the elements - significant industry buy-in to the project, multiple companies investing time, financial and human resources in building on the project.

In the comments I argue that OpenOffice.org should consider setting up a 501(c)6 (a trade association) as Eclipse did, to ensure both community and industry participation in the project:

OOo as a project is really too big to be easily accessible to a volunteer community, but the project has succeeded in gaining industry support - an initial board would doubtless include IBM, Sun and Novell as major members, but might also include CollabNet, the French ministry for the interior, maybe NeoOffice and StarXpert?

In any case, the structure of a trade organisation, which aims more to have an ecosystem than a wide-open community, seems more appropriate for a project like OOo. It provides all of the things which Michael Meeks has been calling for - an independant governing body which owns trademarks and copyright, and is answerable to companies and communities in proportion to their contributions.

I also think that it’s important to separate governance in the sense of marketing, infrastructure and industry relations from technical governance. In the case of the Eclipse Foundation, it’s important to note that IBM is still by far the greatest single contributor of code:

And while it’s absolutely correct to laud praise on IBM for Eclipse, it’s worth noting that even now, 7 years after the project has been freed and 5 years after the creation of the Eclipse Foundation, 75% of the committers work for IBM, and an even higher percentage of the check-ins come from IBM employees. So yes, the project has succeeded in establishing an independent governing body, but code talks, and IBM still talks loudest.

Aside from that, I want to reply in particular to something that Ted said in his comment:

Community governance is hard? I’m going to have to call bullshit on that. It really isn’t hard. What’s hard is letting go of control, which Sun has proven to have an extremely hard time doing.

I agree - letting go of control is hard. And I’ve seen many companies struggle with it - Xara, Wengo, Sun, to name a few, and other companies skirt the issue by unashamedly keeping control - Trolltech, MySQL, Alfresco, JBoss, SugarCRM come to mind. It’s a question of expectations. When a company says “sure, we’re happy to work with you, on our terms”, you know where you stand.

But starting a project on Sourceforge, putting 4 years worth of code on there, telling your team of (proprietary) software developers “now you commit there”, and then expecting that Poof! like magic little Code Gnomes start appearing from out of nowhere to make your project better is unrealistic. It really is the difference between “organic” (grown from scratch, by developers for developers) and “non-organic” (code is liberated en masse) projects. If you have absolutely no governance guidelines whatsoever, who’s the maintainer? The manager who manage[ds] the development team in your lab? How well does that work?

Upgrading to Ubuntu Hardy: a typical dist-upgrade, so far

I upgraded to Ubuntu 8.04 from 7.10 today - I set the upgrade going when I went away for lunch, half hoping that it would be done when I got back. So here’s my experiences so far.

•  The upgrade stopped with a question screen after about 15 minutes. The installer wanted to know if I wanted to replace one config file which I have never touched with the distribution’s version.
• After that, the install blocked a further 9 times, one time for an OpenVPN password, when I would have much rathered it carry on without starting the VPN, and eight further times for config files. I had only changed one of these files since my previous upgrade, and would have liked that file to be kept without a question being asked.
• After rebooting, my screen was filled with error messages from crashing applet, many of whom have not been installed on my panel since I upgraded from 7.04 (because they didn’t work with 7.10).
• Apport was nice enough to offer that I create bugs for each one, which I tried to do, but apparently there was a problem with Firefox after the upgrade due to a release I’d previously installed separately, so that didn’t work until I restarted Firefox, at which point it worked swimmingly. Some bugs reported in Launchpad, but I really lost track of where I was at.
• Upgrading Gossip lost old account information. Apparently the DTD for accounts.xml changed, and the new version of Gossip can no longer parse the old accounts file. Bug reported.
• The Xrandr applet works again, after being broken in 7.10. Nice.
• Dasher still crashes when changing language or dictionary, or when importing training text. Hitting F1 in Dasher does nothing. The Dasher manual installed doesn’t correspond to the Dasher user interface. Bugs reported.
• Suspend/resume gave me a black screen the first time. I know stuff is happening when I open the lid; the wifi indicator shows that I have network, the hard drive light is flickering, but I have no screen. I’m hoping it’s a one-off, and that it’ll work now.

All in all, not what I’d come to expect from Ubuntu, although not an unfamiliar experience for me over the years. Perhaps a straight install would work better than a second dist-upgrade on a system that has actually been lived in. I haven’t tried everything yet, obviously, and I’m looking forward to seeing if there are any improvements in the support of my webcam’s driver - although I’m not holding out much hope.

Trademarks

Funny, this post has been in my drafts for months… in relation to my earlier post, and since a trademark issue is at the heart of much of the recent OpenSolaris controversy, the time felt right to finish & publish it.

Many moons ago, there was a discussion on the FLOSS foundations mailing list about trademarks for the Nth time, after Simon Phipps proposed having a BOF on the subject at OSCON.

My initial reaction was “I hope that people find something new to talk about”, I’ve been involved in many conversations on the application of trademark law to free software projects, and typically, the range of reactions is:

• Defending trademarks is important, and the (US) law requires aggressive defense (the Mozilla or Wikipedia position).
• Defending trademarks is important, and we can draft guidelines which allow some community uses of the trademark, but we have to disallow a wide range of things to avoid opening a loophole for malicious use (the GNOME position - the degree to which we’ve succeeded is debatable - or the Perl Foundation).
• Defending our community is important, but that doesn’t require a trademark (the Postgres position, or Chris Messina’s community mark idea)

There are lots of data points between all of these (Linux, Open Source, Eclipse, Java, …) which go from the “we didn’t register the mark, and we regret it” which perhaps apply to Linux and Open Source, to “our trademark is a certification mark” for Java. I would say the most common reaction is “we have to register the trademarks! But we have no idea why, or what that means for the project.”

Read the rest…

Sun: Trying to do the right thing

I’ve been annoyed by some of the Sun-bashing that has been going on over the past few months and years. I’ve written in the past about my belief that Sun are trying to do the right thing, and my appreciation for the investment that they’ve put into projects I care about. And yet no matter what they do, it seems like there are nay-sayers working to undermine Sun’s community-building efforts at every turn.

Here’s a few examples of Sun-bashing that I’ve seen recently:

• No projects primarily sponsored by Sun get accepted to the Google Summer of Code (unless you count MySQL). Rumour has it that Sun were told not to bother applying. Of course the Summer of Code is Google’s baby, and as such they decide who gets to participate and who doesn’t. They don’t even have to explain themselves.
• Linux Foundation employees repeatedly criticising OpenSolaris and Sun. I suppose that this is to be expected from a group that is representing its members, and sees the OpenSolaris kernel as direct competition to the Linux kernel, but it’s just as disappointing to me as when I see KDE or GNOME hackers ripping into each other
• Press articles in Slashdot [2] [3] and elsewhere consistently spinning things as “Sun’s free software efforts aren’t sincere” interspersed with “Sun is ruining <insert project here>”.

I feel like a lot of this rhetoric is self-fulfilling prophecy. If you say often enough “Sun is a bad community player”, then Sun’s projects will seem unattractive to prospective volunteers.

All of this completely ignores the many great free software people who are working for Sun - to name just a few, Glynn Foster, Simon Phipps, Dalibor Topic, Ian Murdoch, Rich Burridge. These people are extremely clueful about free software and community interests. And the message which we have seen consistently from Jonathan Schwarz over the past couple of years reinforces that there is a commitment to free, community developed software, and there are many capable people working towards that commitment within Sun.

So why the difficulties? Many of them, I think, are project specific, and stem from this fundamental fact:

Community governance is hard.

Read the rest…

Red Hat, Novell, Canonical and the free software desktop

Lots of people are up in arms because Red Hat’s desktop team released a statement containing this: “we have no plans to create a traditional desktop product for the consumer market in the foreseeable future”, and Ron Hovsepian said “Novell’s Suse Linux at the desktop is unlikely to be popular with consumers in the next three to five years”. To me, this is not defeatism, it is simply an example of positioning in action. Last year at Solutions Linux in Paris, I did a little experiment, designed to show that Mandriva have a problem with their positioning. I asked several people to tell me what market they thought the following popular distributions targeted:

• Red Hat
• Novell
• Ubuntu
• Mandriva

The answers were unanimous:

• Red Hat: Enterprise servers
• Novell: Enterprise desktops
• Ubuntu: Consumer desktops
• Mandriva: Ummm…

Read the rest…

Famous people supporting the Libre Graphics Meeting

…well, for some definition of famous (and certainly mine).

Without wanting to play favourites (I love all our donors) here are some names that stood out when I was reading through the donors list earlier: Michael Tiemann of OSI and Red Hat, Bdale Garbee, everyone’s favourite gentle giant from SPI and HP, Garrett Lesage and Andreas Nilssen from the Tango project and more, Stephen O’Grady of RedMonk, Ilan Rabinovitch, chief bottle-washer of SCALE, Nicolas Spalinger of the Open Font Project, Chris Messina, the man behind the original incarnation of SpreadFirefox, John Bintz, who gave a great presentation last year on using the GIMP and Inkscape for comic drawing, Andy Fitzsimon, Aussie artist & Inkscaper extraordinaire, Louis Desjardins, last year’s organiser, Eric Sink, future GUADEC keynoter and co-founder of AbiSource, Adam Sweet from the LUGRadio crowd (where are the rest of ye, lads?), Steven Garrity of SilverOrange, best known for his work on GNOME and Firefox, Mark Wielaard of Classpath, and the rocking Jon Phillips of Creative Commons and Inkscape.

Modestly forbids me from mentioning myself (damn!).

If you’ve been holding out until the last minute, notice is served - we are now closing in fast on $10,000 raised, and the campaign ends tomorrow at midnight, PST. I am hoping that we will reach $12,000, which would be a great achievement, even if it does fall short of the original $20,000 goal. To that end, I’ve just changed the campaign goal to be more realistic - let’s see if we can get $2,000 more in the remaining 36 hours.

Update: Jealous of all the attention other famous people were getting, the very famous Vincent Untz jumped in and pushed us over the magical $10,000 level last night. That was quickly followed by donations from the famous Behdad Esfahbod and the famous Paul Cooper. And I don’t know how I failed to mention a friend from ILUG, Padraig Brady, also famous in his own right. Thanks guys! Keep ‘em coming!

Correction (for the record)

I just listened to LUG Radio 5×15, including my interview about LGM this year. And I have to make a correction. I have tried, but I cannot find any way to drag & drop or copy & paste a curve from Inkscape into the GIMP. I can go via the intermediary of an SVG, since importing gradients and curves from an SVG drop or load is supported in the GIMP, but I can’t figure out how to drag & drop elements from the Inkscape canvas into any other application - when I hit the edge of the window, it just starts scrolling. And cut/copy & paste isn’t any more successful.

Any Inkscape people out there able to set me right?

One interesting drag & drop thing I love showing to people is dragging a chart created in Gnumeric into Inkscape - the drop is a proper SVG, and you can ungroup & manipulate individual elements from the chart in your favourite vector graphics application. Which is nice.

Thank you Michael Tiemann!

The Libre Graphics Meeting fundraising campaign has now surpassed $6000, including our first thousand-dollar donation, courtesy of Michael Tiemann, board member of OSI and VP of Open Source Affairs at RedHat.

Thank you Michael, and thank you everyone else who has donated! Our community is now a bigger sponsor than Intel and Google, and is officially our biggest sponsor this year! Only $2000 more, and our community will be a Gold sponsor, and over half way to Cornerstone sponsor level.