Building a successful community-driven software project: part 2
This is a multipart article – link to part 1
This is a multipart article – link to part 1
There's a lot of buzz about open source software lately. Many projects are supposed to embrace the Bazaar development model, but they really don't, and their leaders just seem to ignore users a lot. This often leads both to wild forking – one new feature, one new fork that's never merged in the original code base – and to multiple incomplete implementations – there're many projects to satisfy a single need, but none of them is complete.
What should you do to make your project successful? Those are my guidelines. I've split the post into pieces since it was growing too large for a single entry.
This is mostly developer-oriented; I think supporting a community of end-users might be slightly different, although concerns might sometimes be similar.
1. Make it easy to discuss about your software.
Something might not be clear about your software, whether it's usage, availability, license, scope, and so on. It must be easy to ask a question, whether it's on a mailing list, on a forum or by any other mean – Launchpad Answers or GetSatifaction are good alternatives – just don't make it too hard to just ask a question!
If you're using a custom forum or any other portal, make sure it supports a decentralized authentication system, be it OpenID, Google Account, MSN or whatever you like. Very often creating a new username/password couplet, fill in a few forms, clicking to verify an email, etc. etc. may just be too much an hassle for somebody who just wants to ask a small question.
Similarly, if you're using a mailing list, make sure it's not too hard to get subscribed; if possible, just let anyone post a question – then a “reply all” from any list subscriber should reach the poster. Don't moderate subscriptions and posts unless you've got a really good reason to – e.g. too much spam that automatic filters can't handle. Consider using mailing list providers that make it easy to use the ML from the web as well – Yahoo Groups and Google Groups are great examples; they'll give a great overall user experience, and satisfy those who prefer forums and those who prefer mailing lists at the very same time.
2. Answer questions and doubts.
This seems a silly one, but it isn't. Make sure users get a minimum of feedback, be it just a “read the FAQ”, or “it's not implemented yet” or “I don't know”. Any message left lingering is a big thumbs down for your project.
3. Have a FAQ
Apple has announced it will contribute to the OpenJDK project, so there's probably no risk that MacOS-specific pieces of the current JDK will be lost. Probably the next, Oracle-backed JDK will be just fine on Mac and will probably be released on time.
There're two things I don't understand:
I'm not an hardcore Apple user. My first Apple product ever was a Macbook Pro I bought about one year ago.
But I fell in love with the Mac platform, and I think its really strong point is that it's versatile. It's got plenty of useful proprietary apps that sometimes I want to use or I am forced to use, like Microsoft Office or Adobe Photoshop, a pretty decent Adobe Flash implementation, but since it's a BSD derivative it's got a lot of nifty tools I'm used to and that I really like.
Sure, Windows might be faster at many tasks, especially if you're gonna spend as much money as you need to invest in Apple hardware, but my feelings towards Windows are mixed. I liked XP at the beginning, but it grew old and tired; Vista was a mess; Seven seems fine, but Microsoft still needs to conquer back my trust. Also, Windows is fine as a development platform as long as you use Microsoft tools – which are quite good – but it can get squeaky as soon as you adventure into the FLOSS land – I can remember some madness I incurred in while trying to use git under Windows.
What about Linux? I's a great development environment. I simply love my Ubuntu box, it's exactly as I want it. Eclipse, Netbeans, everything is set up as I want, all the key bindings, all the details. But my Ubuntu box is a tower form factor workstation with ethernet connectivity and two monitors – I think Ubuntu fails pretty hard at the common desktop experience, especially on a notebook – wireless connectivity is often an issue, suspend and hibernate are unreliable, the battery doesn't last as long as I'd like, and Flash on full screen is flaky. I don't think it's Canonical's own fault – it's really hard to have an OS work on each and every hardware and giving it away from free – but from the user's point of view this is a failure. Some key apps are also missing – OpenOffice might not be enough a replacement for Office, and GIMP is not quite as effective as Adobe Photoshop.
But the Mac has it all, and that means it's got a lot of appeal on developers – we usually want the eye candy and the dark console, we like our computer to 'just work' and yet we like to be able to tune every detail. Whenever I go to a conference I can see there're far more Macs, iPads, iPhones than in any other typical business environment. A lot of developers from many communities simply love using Macs for their development work - I simply don't think that developers don't count in Mac market as someone stated. Maybe many of them are not Java developers, but as soon as they're unable to install Eclipse they'll complain as well.
Apple's Java sports its own Cocoa bindings, which are very useful to achieve a native look-and-feel, OS integration and general responsiveness for all java based software. Eclipse is java based and it's a widely used platform, RubyMine is one of the great favourites for Ruby coders, Netbeans is evolving at a great pace and supports Java, Ruby, Python, PHP and many others. Not to mention the fact that while I write those lines it's the only JDK available and working out of the box on Snow Leopard.
Now this is going to change. Java is deprecated on Snow Leopard, won't be included in Lion and can't be used for Mac App Store.
If Mac OS X were to be left with a third-party Java implementation – and let's hope SoyLatte can cope with such burden – we'll probably lose native OS integration, everything that's Swing or SWT based will need X11 to run, Swing will probably use its native Metal widget set while SWT will probably rely on GTK+ bindings; Apple doesn't even seem to be willing to contribute its own patches to the OpenJDK project, even though it would be meaningful if they just didn't want to waste resources on what they think it's a dead technology.
It's not a great scenario, but the real question here is: why?
Why should Apple have done all that? Is it that difficult to update the current JDK to work with Lion? JDK7 might be a different topic, but hey – they're basically dropping what they've already done! And they're not letting anyone to continue their job easily. I think Apple is perfectly aware that the Mac OS X ecosystem is fairly different from the walled garden of IOS – Windows is still got a much larger market, and limiting what Mac OS X can do will just feed Microsoft's pockets again.
I don't think there will be an official statement on that, but I bet Oracle has a part on this. Mac JDK seems to have been built by Apple but to have sprung from Sun's own JDK – is Oracle not willing to let this deal to go on? Maybe Apple needed to renegotiate its license and it turned out to be too expensive? Even though Apple may think Java is dead, it has no apparent reason to silently stab it again.
Their decision is causing a lot of noise in the Mac/Java community, let's see if Steve comments on this as well