Day 3 is now in the books. I tried to go to as many sessions as I could and avoid booth duty. I’m just not much of a fan of the show floor. It’s too loud to talk and people really only want your free swag.
The keynote was just ok. I found Tim O’Reilly very boring. He kept talking about Web 2.0 which I’m guessing is his phrase. He gave a bunch of stats on the sales of books as a measure of language popularity. I’m sure that this is a metric of some value but book sales alone don’t necessarily tell the whole story. Python and ruby sales were up, perl sales down. O’Reilly must have a job board now as well as he talked about requested skills. SQL, Java, and C++ were the top three. It just reinforces my belief that the data is absolutely the most important thing.
Kim Polese presented what was really just a marketing session on SpikeSource. Andrew Morton talked about the Linux kernel. He at least acknowledged that Linux desktop isn’t really happening. He also took a pretty hard dig on OpenSolaris. Jeremy Zawodny from Yahoo talked about how they used open source. It was pretty much the expected laundry list of projects. It wasn’t clear to me if they are still strongly committed to FreeBSD or if they are looking to go to Linux. I got the sense that new servers are all Linux.
The best speaker at the keynote was Jonathan Schwartz. Schwartz got hammered on if Java will be open sourced. He danced around it a bit but gave forking as one reason to keep it closed. Sure, that’s a possibility but that just means the pressure would be on the “official” Sun project to do the best job. And he also said Sun engineers would likely be contributors to Harmony. Feels like they are playing both sides. He also of course talked about OpenSolaris. He talked tough telling Linux to forget the religion now that Solaris is open source and fight it out on technical merits. But is it really just too little too late?
I spent the rest of the day attending sessions. I went to a session by Theo Schlossnagle of OmniTI on Oracle to MySQL replication. The approach they used was pretty similar to the methods we used in SQL Server for Oracle replication. Their approach was simpler and didn’t maintain transactional consistency on the replicas. He claimed this was never really a problem but I find it hard to believe.
I also attended the state of the Linux kernel session. Let’s just say that it is absolutely amazing that Linux even works. The kernel dev model seems so chaotic to me. I guess it works for those guys but I have to say I prefer the FreeBSD model better.
Post a Comment