Someone wasked me today whether I'd read
Domain Driven Design. This is what I said:
I've read part of it and it's on my bedside stand. So far I've agreed emphatically. Rich and real Java domain models.
I think it has some implications for
uPortal. Currently we're modeling a lot with XML, and XML has some nice properties, but we could get a lot of mileage about a richer and more specific domain model for portal concepts of tabs, columns, layouts, etc. Creates the objects that would answer AJAX callbacks, the domain model with which tools and widgets would interact. I think it could be very valuable but I'd like to get some time to look under the hood on other opensource portals and see what they're doing.
I think it could help make the continued holy grails of other layout metaphors and added behaviors more concrete. If you really believe in embedded tree controls or a tab that's really an alias for Sakai content, write the TreeControlledTab or the SakaiTab and drop it in the layout widget bucket.
I'm getting burned out on XML / XSLT. In some ways it's very powerful, but it's also pretty monolithic and unagile. Where we do need to use XML, let's translate a rich domain model to XML. And where we can just use the domain model in, say, a JSP, let's do that. I'm all for going to XML XSLT when it's the best solution to a problem, but let's stop regarding it as the only tool in our toolchest.
Enjoying the book. I should finish it so I can discuss it more intelligently.
In general I tend to like books much more than other developers. You can thumb through it, lend it to someone, throw it in anger -- they're really quite versatile. The keynote at JA-SIG was interesting but I hope books do not go the way of the dinosaur as he predicted anytime soon.
