I got a good night's sleep last night so I'd be fresh and ready for the Smackdown today.
Matthew Porter, Scott Delap and I visited Jive
Software's offices last night and had a great time sipping suds in their
beautiful downtown office.
My AppFuse tutorial yesterday was well-received by a packed room of developers. Rather than writing code the whole time and doing a measly 30-slides, I added a bit more meat about Spring, Hibernate and testing. Most of audience was unfamiliar with Spring, so this seemed like the right thing to do. Of course, this led to more talking and less coding, but most of the folks I talked to were nevertheless very happy with the tutorial. If you'd like, you can download my presentation from the event.
Thanks to Rob Harrop and Thomas Risberg for both letting me know about the lack of Spring experience, as well as sitting in on my session. It was pretty cool having these guys in the room, as well as SiteMesh/jMock inventor Joe Walnes. Without these guys, many of the cool features in AppFuse would not be possible.
Now I'm sitting in the beginning Keynote session at OSCON, where they've announced they have a record 2000 attendees this year. In addition, it looks like OSCON is in Portland for the long run - this is the 3rd year it's been in Portland. Rather than moving to a new city like they used to, they've decided to stay b/c conference attendees like it so much.
The "unofficial" tagline of the conference is fun. Open source is fun and exciting - both to develop and use. This is in stark contrast to closed source software that tries to stay stable and boring, with no surprises.
O'Reilly is not just book publisher or conference producer - but also a company that looks to the future and tries to figure out what's next. To highlight this vision, they've created O'Reilly Radar.
We're currently going through "The Open Source Paradigm Shift". Integration of commodity components has led to a new model where value gets captured. Rather than being at the software level, it's at the services level.
Key Questions for Open Source Advocates
- Will "web 2.0" be an open system? What do "open services" look like?
- Data as the "Intel Inside" - will we end up needing a Free Data Foundation in 2010?
- How does the paradigm shift change our business models and development practices?
- Who shoujld we be watching and learning from?
Things on O'Reilly's Radar
- Ruby on Rails: new platform and new language. May well be the Perl of Web 2.0.
- GreaseMonkey: a Firefox extension that alters websites to fit your view. A website is traditionally closed. GreaseMonkey "opens up" a website and rewrites it for the user.
- HousingMaps.com: leveraging Google Maps and existing data from a bunch of different webservices to build a better website.
- Findory: Uses the articles you like in blogs and news, and finds similar articles. Similar to Amazon's recommendation system.
- Internet Telephony: Asterisk in particular. Skype and Broadvoice. Broadvoice is pushing BYOD (Bring Your Own Device).
- HomeBrew: similar to Tivo.
- Opening up hardware, not just software - i.e. Car PC Hacks, Smart Home Hacks.
For more, go to the Visualizing Technology Trends on Thursday afternoon. For the first time, the computer book market has stabilized. This is a good sign that the computer industry is about to start rebounding. As far as the book market goes, it's market share and growth - Java still leads the pack by a pretty wide margin. A large reason of this is due to Open Source Java. SimplyHired.com spiders 7.9M jobs on 755 different job boards. General books on Linux are up, especially non-RedHat distros. Books on RedHat have decreased significantly.
This is an interesting conference to be at when you're a Java developer. For the most part, everyone seems to be Perl fans, followed by Python, and a few Ruby guys. Most of these developers are very vocal about the fact that they don't like Java. Then again, Java is the leader in many areas - and it's the open source way to hate the guy on top.
Kim Polese, SpikeSource
Building on the Architecture of Participation. A transformation from Do It Yourself (DIY) to Do It Together (DIT). Thanks to the architecture of participation, open source has achieved World Domination - as evident by governments mandating it and IBM pouring billions into it.
The architecture is characterized by:
- Commoditization of software
- Network-enabled collaboration
- Software customizability
In Phase I, we built and we built with. Open source had DIY origins. Now we're in Phase II, where increasingly the action is out in the long tail. Countless new building materials are piling up on the long tail. Now it's possible to build just about anything with anything. IT shops are building a phenomenal set of DIY "packages" that combine components from both ends of the curve.
The two problems with this:
- Velocity mismatch: all components are on different release schedules. Linux, Apache, MySQL - all on a different release schedule. In addition, the ones on the other side (Lucene, Struts, Mambo) are on a different cycle.
- Dependencies: When one version of one product changes - what happens to all the dependencies?
To solve these problems, companies are developing formalized proceses like review boards, support centers, OSS incubation centers, testing groups and they're certifiying / defining stacks internally. Most of this work is laborius and not related to the core competency of the business.
What's next? Phase III - IT becomes core. They do this by offloading critical but non-strategic work to independent service companies. DIY evolves into DIT with the help of independent service companies. Of course, this is all leading up to the fact that SpikeSource provides these services. It's funny that as soon as Kim said "SpikeSource" - all the presentation screens in the room quit working (not on purpose). A minute later they're back. This goes to show that marketing is not liked by the Open Source Gods.
"Testing is the single biggest refactoring shift in sofware." - J.P. Rangaswami, CIO, DrKW.
We need testing on a massive skale. For this reason, Murugan Pal and Ray Lane started SpikeSource. They saw the next phase is testing open source software so we can scale testing, together. Solve velocity mismatch and dependency problems with rapid per-defect patch management and dynamic stack configuration.
Testing has always been the software's ugly stepchild. We need to scale open source testing the way we scaled open source development. Some perspective: Microsoft has a 1:1 ratio of developers to QA Engineers. There's no Microsoft for open source software, nor should there be. To solve testing on a massive scale, you need participation plus automation. For models of how both scale, think eBay, Google and Amazon. Their best assets were their customers that supplied data that made their services more useful.
Testing is just one service among many. The Linux distros and middleware core building blocks have been there for awhile. Now we have applications and service companies as well. Who benefits when we have abundant integrated, tested, validated automatically patched stacks? IT and ISVs shift high-value development resources to customer-faces - differentiating features and services. In addition, many other groups benefit and higher quality software gets developed.
Testing will do for open source what it did for chip design a generation earlier. Testing is what catapulted the chip industry forward in the 80s. The new testing tools moved VLSI foward. Countless new IC-powered products were made possible and at much faster development speeds. Solving the testing problem can't be done by one company alone. "Come Test with Us..."
After Kim, another speaker (Andrew from OSDL) began his talk. He talked in a monotone and lacked a presentation. The room quickly began to leak people, me being one of them.