All entries for Thursday 01 July 2004
July 01, 2004
I've been to two talks on Aspect Oriented Programming, one broad on whether AOP is ultimately good or evil, and one (very good) on how metadata might be used by AOP systems. By and large, what I thought about AOP was confirmed (still difficult, and with some dangers) but the discussion about metadata got me more interested. ISTM there may well be the potential for a much more lightweight kind of AOP than that provided by heavyweight tools such as AspectJ. (To an extent this is reiterating my earlier comment about annotation.)
I've also been to sessions on EJB3, JDO2 and using the current EJB/J2EE/JDO together. The politics is getting fascinating. We had a lively and interesting debate with senior people in Sun after the JDO2 BOF last night (lasting till well after midnight, longer than the BOF itself in fact) about how they are going to resolve the potential conflict between JDO and the announced EJB intention to enter the POJO-persistence arena.
For those who remember being taught by Robin Roos last year, I ran into him at the BOF. As of this Monday, he's no longer an independent trainer/consultant but now works for one of the JDO vendors (not Libelis).
The open source debate has been raging here too, in and out of the sessions, fueled by Sun's earlier announcements. Personally I don't think Sun should open-source the Java platform. OS projects tend to trade off stability for greater innovation, but I don't think innovation is lacking in the development of Java, and I'd hate to lose the stability side of it. Interestingly though, many people seem to be challenging the assertion that Java is write once, run anywhere. Too much still needs to change between platforms. I think JSF is an important step forward in addressing that.
I've been going to quite a few sessions on Java Server Faces (I still have one to go, tomorrow). It's certainly one of the hot issues of the conference. They've ranged from the interesting and engaging to the boring and dreadfully badly presented. One of the best was David Geary's session, which included best practices, client-side validation a la Struts, and most intriguingly, using Tiles with JSF. The only real trick is to stick tiles-scoped variables into request scope so that they can be used in the JSF EL.
David did call JSF an MVC framework; there seems to be no clear line on this. I've also heard differing opinions as to why the EL uses # instead of $, though it certainly seems to be involved with the stage at which it's evaluated, and there was a definite-sounding commitment to merge the two varieties somewhere down the line.
The picture seems to be clear that JSF will take over from Struts; it has achieved commercial backing remarkably fast and within a few month there will be an explosion of tools that use it and components and renderkits for it (Oracle will be releasing 70+ components for it next month). Apart from simplifications over Struts and the fact that it is a standard, the "toolability" (yes, I've heard this word used!) of JSF is seen as the main reason it will succeed.
Now it's almost 10.30pm and so time to go to the JDO 2 BOF.