My current blog software is based on a framework I wrote for my current employer and since I am leaving this month I will have to start over. So I will do the radical (maybe) thing and build it entirely as an Ajax web application.

I wrote a customer web application framework at my job since we have some unique requirements and it seemed logical to write a new one. It was based on ideas in Wicket and shared a lot of concepts but was built to run only in a stickyless web server farm (ie no application session state or synchronization other than a replicated database). Unfortunately the rest of the team never caught up to me and I've sat around for two months being unable to move forward with the actual site since the database and html was never ready. I wound up writing the blog software to exercise the framework (at home to be fair) but since my last day is Sept 29, I can't use it anymore.

My last application at my previous employer was a single html page which was the entire application, all run via DWR with the "pages" dynamically created inside that one page. I did this as an experiment, and it worked really well. They did force me to declare it as a jsp page for (don't ask) political reasons. It used no framework other than some basic java housekeeping.

So this time around I will do something similar, although I may have a bit more "framework" code since I will be supporting multiple blogs with some variations in each, so it might have a more CMS feel to it. I still intend to use H2 as the database and keep the same URLs and structure but stick to the plan of keeping the "business logic" on the server side and generating the "presentation" on the client dynamically.

I am not sure which of the bazillion ajax/javascript frameworks to use besides DWR; there are way too many.

It's doubtful if this works if I will release the blog app (I'm calling it BlogFiche) as open source, as there are too many blogs, CMS, ajaxy things available and everyone is getting glassy-eyed from picking.