Technology landscape

I have not been a great internet buff... Primarily used to surf a bit to get some work done and that is about it. For some time (which is a couple of years) I am trying to make better inroads and right now (which is few months) I really want to be net-savvy and stuff like that... Obvious reasons for that is my need to increase the breadth of knowledge in the technology space which is bread butter and more...
But this has not been easy... especially with the technology landscape being already very very large ... and it is CHANGING... the rate I am clueless but sometime it looks to be light speed...

I am trying to decide on how I can keep afloat and still know sufficient about what is happening out there. One thing which I thought could be to pick up a set of good sites and just follow them and forget the rest. Some sites which I plan to follow (being a java guy) regularly are
http://www.theserverside.com/
http://www.onjava.com/
http://www.javaworld.com/
http://www.ibm.com/developerworks/?ca=drs-tp5007
http://dev2dev.bea.com/
http://www.devx.com/

I would really like to add some SOA and Java Web framework related sites, but don't have any recommended ones. Friends add on to this if you think there are better places to go and better things to learn.
Or for that matter let me know your thoughts on the approach itself. May be or surely there is better way to achieve what I want to achieve. Go ahead and hit me

JSF - My experiences

For the last two-three months I have been working on JSF (My Faces) with facelets. The experience has been good for most parts. The components available out of the box with JSF implementations (standard, tomahawk, richfaces etc.) have been of great help with development. The decision of using facelets was a good one. It helped to make the view much easier to format for user experience. With all these good things, I found some really seemingly trivial but irritating issues.
- Handling blanks with input fields that are bound to non-string values is a problem (though luckily solutions are available - specific converters).
- There were scenarios where I switched from tomahawk components to standard components for making things work. The reason for which I never could understand.
- A navigation rule syntax which is really very verbose.
- A backing bean which can get quiet large and unwieldy depending on the features to be handled.
- Making pop-up screens for lookups etc. - I could never get it working.
Some of these can get irksome, but I still felt that with Facelets, JSF makes web development far more easier.
I would like the JSF gurus to throw light on how these problems can be handled and also add on to the list of the advantages and issues of JSF. It would be of great help to me and many others

Annotations - Is it a compromise between convenience and flexibility

Annotations have been a craze for sometime now. I heard it the first time when I went through an article on EJB 3.0. As soon as I saw it, I was reminded of Weblogic's xdoclet method of allowing the deployment descriptor attributes to be defined in the code itself. And I did not like it at all. I still don't like it much.

I probably am from the old school but I felt that by moving these back to the code, I lose the flexibility the concept got me at the first place. I always had to change the source code if had to tweek any of these attributes. Did not feel right. But I left it at that.

Then I came across hibernate annotations and used them in a project. Was very handy and friendly. I did not have to go open up another file to define the mappings. Cool... I felt as a developer. Still that earlier thought did not go away. But I was able to convince myself that these mapping probably wont have to change much once it is done. Changing table column names once in production anyway would have been taboo...

Today however I heard about Spring 2.5 coming up with annotation based DI and the question popped up again. If I use annotation and declare my dependency right on the class source code, am I not coupling it with the concrete dependency... Is it defeating the main purpose of DI I was not sure...

Can't better IDEs solve the xml menace instead of going for annotations? It is a thought. The industry will decide...