DivaBlog

My Notes on what was said (your mileage may vary and apologies for no links and potential misspellings). -Danese

Tim asked how many Java developers at JavaONE were working in Open Source (because he believes that all the interesting stuff in Java is happening in Open Source, like Struts for instance) and he got not many hands. Of course here at OSCon all the developers who are interested in Java are also interesting in Open Source. Wants this panel to explore questions of whether the status quo way Java has been handled is NOT working...

Eric Raymond (ESR) threw down the fundamental idea of Open Source in the Cathedral and the Bazaar. Bruno Souza is an open source developer, but also the leader of the largest Java User Group in the world (in Brazil). Simon Phipps used to push Java at IBM and then went to work for Sun.

ESR says Open Source developers don't trust tools that someone else has an exclusive lock on. This is the result of bitter past experience. Eric says the right to fork is like the right to sue, the right to strike or the right to bear arms. You many not want to exercise them, but once someone tries to take one of them away from you.

(Tim): So Simon...is Java open enough?

(ESR): No!

Simon Phipps agrees that the problem is trusting the tools. In 1995 Sun was trying to solve that same problem by creating Java. Java (and its emphasis on compatibility) is one attempt at solving the need for "freedom" from vendor lock-in by forcing all vendors to follow an agreed-upon standard. Open Source is another attempt at solving the vendor lock-in problem (via the "freedom" of code availability). So this debate boils down to a need to bridge between two large and successful communities of software freedom. Just as the word "Free" is not enough to describe both "gratis" and "libre", the word "Java" isn't enough. There are many parts of Java and some of them are already "Free" in the Libre sense. Java isn't open enough in the sense that although it is now possible for a compatible open Java to exist, there isn't one yet.

(Tim): So, Bruno...is Java open enough?

Well, in Brazil participation on the JCP lets Java developers specify what Sun and others implement. This is working for for now, but Bruno believes there needs to be an Open Source implementation eventually and that JCP 2.5 allows for this. He says his friends in Brazil would be willing to participate on an open source implementation.

(Tim): Audience...is Java open enough?

Brian Behlendorf...explaining the Apache position says that the JSPA version associated wtih JCP 2.5 (which Apache signed) allows for open source Java compatible re-implementations. The Apache movement sees incompatibility as a bug to be fixed but not necessarily regulated by legal means. Brian says Sun promised (circa JCP 2.5) that Apache would not be required to pass on compatibility requirements downstream. Recently Apache has noticed that the TCK license has a clause in it that doesn't work with the Open Source Definition...that code created in open source at Apache can't be reused without compatibility (required passdown of compatibility). He says that Sun and Apache are negotiating on this issue now and hopefully it will be resolved.

(Tim): Simon, do you want to comment?

This is an artifact of the fact that Sun is dealing with TWO paradigms of creating software...ONE that is bounded by competitors who want to destroy the value proposition by advantaging their implementation against all others, and SECOND from developers who want Sun to "give up control".

(Brian): Microsoft already created a fork...its called C#.

(Tim): Was surprised by number of developers at JavaONE who were cheering for Sun's stewardship of compatibility.

(Brian): Maybe we the F/OSS community need to get the word out how F/OSS supports compatibility.

(Eric): In the F/OSS world, we have languages that don't fork because its in nobody's interest to fork. Why doesn't that convinced Sun that the same can happen for Java?

(Simon): Because the F/OSS community isn't the problem...its the substantial marketshare of big companies who ARE incented to fork.

(Bruno): Notice that IBM sells WebSphere, not Java Application Servers. It is true that we've seen them provide SWT outside of the JCP system. When you hear IBM telling customers that they can ship a JVM without SWING which will be faster, that's a problem for WORA.

(Simon): Because programmers need WORA to work...they need everything there so they know what they can depend on it.

More Audience Questions...

(Russ Nelson): Why isn't there a good Java interpretor in OSS?

(Eric): There is! The problem is the libraries.

(Russ Nelson): Then we should dig in and write them!

(Bruno): Agrees...Its all possible for us to do now. We as a community need to go off and do that now.

(Brian): What we learned at Apache is that to implement the spec, you need to read the spec. To read the spec you need to sign up to the copyright notice (which is bounded by the JSPA). Those restrictions keep companies from committing resources to Open Java implementations and THAT is something the F/OSS community would need to hope to do a compatible re-implementation. Apache is working to try to fix this one problem licensing problem, which they hope will open the floodgates.

(Simon): Predicts that there will BE open source Java (possibly based on Cafe, GNU Classpath) real soon now.

(Tim): Has never understood why Sun gets hung up on theorhetical worries and loses all credit for the much good work they have really done.

(Audience member Scott Dietzen) : Would like to propose a middle ground based on control of the Java Namespace and Brand. (Scott works for BEA and this is their current proposal) to the JCP

(Brian): Apache have also proposed this to Sun. Nothing would give you the right to use Sun's trademarks without their approval in this proposal.

(ESR): That would be compatible for OSD.

(Simon): This is worth pursuing (in his opinion, but he isn't going to speak for Sun on this).

(Bruno): Feels this type of debate is important because there are two communities here who are seemingly at odds...There ARE Java developers who are worried that Java isn't open source yet. Sun needs to hear this.

(Tim): But maybe also the F/OSS community needs to hear that there IS evidently support in the Java community for Sun's point of view among their key constituents. So there needs to be more dialog...How can we get this discussion to happen more constructively?

(Audience Question): How can we help?

(Bruno): Download the F/OSS version of Java and use it...then contribute to enhance it!

(Simon): Programmers join libraries together these days, so VMs are sort of irrelevant. There is still time to work out how to get Java open.

(Audience Question): Is Mono having any implications for F/OSS server-side in the future?

(Tim): Seems to be Java all the way on the server.

(Bruno): There is a BOF tonight at 8:00 about this...come and lets see if we can go anything more ;-)

My Notes on what was said (your mileage may vary and apologies for no links and potential misspellings). -Danese

Tim notes that OSCOon attendance is growing (again) perhaps its an indicator the industry is coming back (applause)!

Tim says O'Reilly doesn't publish books...they try to capture knowledge from the people who are inventing the future. Sometimes that's a book, sometimes its a conference and sometimes its pure activism. "The future is here, its just not evenly distributed yet" - Gibson. For example, CarPC Hacks (an upcoming book) describes what some people are doing now in custom car mods (which we expect to see coming from Detroit in 10 years).

OReilly has been publishing more books about Google and Yahoo and other web implementations. He used to say at OSCon that OSS should want not to replace Microsoft, but rather to become the "Intel Inside" of the leading edge. Clay Christiansen, who wrote the Innovator's Dilemma, talks about commodization of software as a migration of value to different levels.

History of personal computing a la Christiansen...His law of "Conservation of Attractive Profits" meant the API lock-in at the software level alllowed hardware to be commoditized. Open Source tried to break the software lock-in, but Tim thinks all we enabled was a lock-in above the data level (data is proprietary). Lock-in was achieved by "Network Effect" meaning that the massively connected market only needed one provider of each service.

Tim thinks the Internet is the platform now. Apps are built on top of open source underpinnings, but are not themselves open source. Open Source (as we know it now) is less and less a solution. Source + Compliation != Application anymore. You could give someone all the source for Google, but you'd still have to run 100K servers to match the Google experience. O'Reilly has been pushing to have Google and Amazon open up APIs, but Network Effect makes it darned hard to level the playing field.

Getting your users involved in the product is key (Amazon gets its users to contribute content and that makes them feel ownership of the tool and enhances the value....MapQuest is commodity software on top of a commodity database and according to Tim they've stopped growing and are being copied. Tim says if they'd figured out how to get their users to make the service richer they would have dominated. Navteq is a company Tim is watching that IS working on getting users to add value (they are about to IPO - wanna bet Tim has fnf?)

So, takeaways...Data is becoming the new frontier. Start thinking about how to build a participatory layer around the data.

-Free and Open Source doesn't guarantee freedom when applications depend on network effets and data lock in more than software secrecy.

-It is key to invite users to build services and data, not just code.

-Think beyond Linux. Its the whole stack!

-Open Questions: Who will control the data in the future, Who will control the namespaces? Who will win the stack integration wars?

Switching Topics, Tim showed Microsoft's connection tool (called Wallop) that give a visual picture of the social network. Suppose Google put GMail and Orkut together? They would control the data, which is a huge deal now.

Call to Action!!! F/OSS needs to "Napsterize the Data"! We need to own our own data at least. We need to think about layers of access (private contacts, public contacts). The Network is the fundamental platform for applications. Think about Federated Identity Systems. Look at what Apple is doing with iSync (its a huge potential control point for them). Look at MoveOn.org, which no matter what your politics might be must be recognized as a leading edge of the Napsterization of Democracy (expressing the voice of the people).

Switching topics (again) Sofrtware Above the Level of a Single Device - (quote by Dave Stutz)

Look at iTunes and the iPod. Its seamless integration. The music store web backend and the small device iPod are elegant. A system designed from web to hand. Rendezvous (oops, ZeroConf) is cool! (Note that Tim really hates the new name, OpenTalk). Rendezvous Tutorial yesterday yeilded a Ruby port. Keep an eye on this!

Another interesting thing...Nat Friedman's Dashboard. Written in Mono, Nat has the idea of "CluePackets" that enrich your working environment

Firefox is finally revitalizing the Browser Wars.

Mobile (Tim showed MobileWhack blog for example) is interesting...watch this space (noteworthy mention of Python on phones)

O'Reilly Safari is an something Tim hopes you'll look at. They are annoucing an affiliate program for Safari, giving commissions like Amazon does for selling accounts, exploiting the new Web Services API (you can search Safari from inside Eclipse). They are adding special services for academics. Generally they are trying to make the Safari experience more particpatory. Think this is a great way to do Corporate documentation. Want also to talk to people who are working in developing countries (contact Tim if you have any ideas).

Network-enabled Market Research. He told the story of asking Gartner and IDC about Open Source and got back the answer, "Our customers aren't asking for it yet". What kind of Market Research waits for customers to mention trends? Tim showed a visual map of the book market. Java is the biggest topic in the tech book market, but notice Ruby is growing as a topic. In terms of book sales...Java, Visual Basic trending down, pHp rising, also C# rising. He showed visualization of the market by locale as well. Interesting to see who is buying/reading what -> where!

Finally an announcement that O'Reilly will be co-hosting the next MySQL Developer's Conference next April