I'm contemplating the week @ JavaOne while I wait for my flight at the airport.
AJAX/REST
AJAX was obviously a popular topic at the conference. There is a clear consensus that AJAX is cool, gives you the appearance of rich client capabilities, but is still hard to do well. A number of speakers pointed to bugs in various Google products. The point being, is that the Google brain trust does AJAX very well and even they run into problems. JavaScript features are uneven across browsers, so this has lead to a wealth of vendor server side "Frameworks". Frameworks that try and solve this problem. There seems to be dissagreement as to weather or not this is the way to go... See the JackBe/Sun discussion.
Representation State Transfer or REST popped up in a number of talks... Simplicity and leverage of the standard Web infrastructure are it's primary benefits, but a big drawback is that it is not discoverable... although there is work being done in this area.
XML
XML processing continues to receive improved support..... Mark Reinhold gave a brain dump of his thoughts on how XML can be made "native" to Java in Dolphin.
Back To Basics....
JEE - The Java Enterprise Edition version 5 has made good use of Tigers annotation capabilities and lessons learned from the Spring Framework and simplified the EJB specification considerable. Some would still say it wasn't enough and that Spring's POJO model is still the way to go.
SOA - Although vendors continue to introduce products that support all the latest standards and clain "Buzzword" compliance, there is an undercurrent of chatter from attendees that "things" are overly complex... The S in SOAP stands for Simple, but anyone who has read the WS-* specifications can attest to the fact that the Web Services space seems to be spiraling out of control. There were a number of well attended talks in the REST space and mention of WOA as a simpler more scalable alternative to SOA.
Participation
Finally, there was a continuous call for participation of individuals in the JCP. Doug Lea, spec lead for JSR 166, was called out as an excellent example of how individuals can have large impact on the language and platform.
