Great first day at OSCON. Here are some notes from the sessions I attended Wednesday…
Tim O’Reilly
Tim gave his standard Web2.0 pitch in one slide and promised it was the last time. He also mentioned OReilly Radar and I finally got the reference to M*A*S*H. It was good to see that Java book sales have rebounded from his prediction last year that book sales showed Java was dying.
Kim Polese
Nice riff on on Doc’s DIY theme with DIT (do it together). Kim describes things very much the same way that we do at Optaros. She points to the “productization gap” that is the difference between commercial products (and client expectations) and OSS projects. Effectively, companies need to be responsible for productizing (and integrating) their own OSS stacks. And the 2 main problems are resolving the dependencies as the number of components grows and dealing with the “release velocity mismatch” that occurs with separate projects.
Kim points out that companies like SpikeSource (and Optaros) can help deal with these issues.
Andrew Morton
Andrew was very pessimistic on the timeline for desktop Linux. He predicts desktop Linux will show up first on point-of-sale machines and locked-down systems.
Andrew also made the point that the user-to-developer communication lags developer-to-user communication. There needs to be a better way to get user feedback to the developers.
Jeremy Zawodny
Jeremy made the interesting assertion that Yahoo! naturally looks to OSS because any app they build might take off massively. Yahoo! trusts OSS to be able to scale to the levels that they hit for things like this. To me, this really flies in the face of companies who winge about whether or not OSS can scale. When a company like Yahoo! has faith that OSS can scale you need to wonder what other companies are thinking.
Jeremy also made the point that Yahoo! uses many (10’s, 100’s?) OSS projects. When you look at a list of 10, 20, 50, or more OSS apps you quickly realize that the company using those apps can NOT be asking for “one throat to choke”.
Jonathon Schwartz
Nat grilled Jonathon pretty well about why Sun won’t open source Java and why they chose a license for OpenSolaris that won’t comingle with the GPL. For Java, Schwartz played the “fork card” and raised the specter of a world where Java forks broke compatibility. In the next breath he mentioned that he’d be thrilled to see a thousand distros of Open Solaris.
Jonathon made a very interesting assertion that “every project at Sun will (eventually) be free and/or open source”. This is because the value has moved from the software to other things like update-networks, advertising, etc. JS then got a bit fast and loose jumping between “free” and “freedoms” but still this is a pretty bold statement from a software company.
Peaceful coexistence panel
This was an awkward panel. No one really wanted to defend the proprietary software point-of-view so it felt like the “debate” was just against a devil’s advocate. At any point, there were some interesting comments made.
Woods- Companies need to gain the skills to overcome the productization gap. I’m starting to feel this is true more and more after reading things like the “Open Letter to OSS”.
WebFrameworks
Got to see Craig McC present on the similarities of the diff frameworks. I was hoping this would be more future-looking but it still nicely laid out a way to compare them. One big suprise for me was how few attendees have used anything except Struts. There were only a handful of people who admitted to using the other frameworks like WebWork and SpringMVC.
Semasiology
R0ml got a good turnout for part 2 of his presentation and left the audience clamoring for a part 3 next year.
OpenData
R0ml presented his idea for a SchemaForge where people could collaborate on RDBMS schemas the way they colloborate on code at SourceForge. The audience seemed pretty interested in how this would work so hopefully he can get something going with his newly registered domain.
ID Mgmt BOF
I tried to anonymously sit in on the Identiy BOF room to catch up on email and get a sense of what the ID gang is concerned about. But the group made everyone do introductions. So I obfuscated my identity. How is that for irony. I was only partially paying attention so I may have mis-heard this but I believe I heard the phrases “feel up the elephant” or “light up the elephant” in reference to the blind men and the elephant allegory.
I’m pretty sure this was brought up, but what were his arguments against XML Schema for schemaforge? I agree–that is a terrific idea of a *forge for data models, but wouldn’t pointy brackets make more sense?
At any rate, I brought up this point for internal use at my company. Much as you can look at object models with the 4 archetypes in mind, I think data modeling falls largely into that line or pattern of generalization as well… A versioned repository of standard data models is essential as service oriented architectures mature and systems become increasingly interconnected.
The ideas of XML Schemas and Microformats was raised and the reason for focusing on RDBMS schemas is that the DB is where the data is. I don’t have statistics to back it up but most structured application data is stored in databases. A few enterprise apps, such as some CMS systems store data in XML but most apps store it in a DB. So databases would be the first place to start to tackle the problem.