Shay Shmeltzer, the lead of JDeveloper at Oracle, just blogged about the relative sizes of NetBeans and JDeveloper communities. Good sign! Assuming that the European Union won't do enough damage to destroy the Oracle / Sun deal, it could be the beginning of a talk between the two communities.
But I don't agree with Shay's view about the size of the two communities. In particular, I think that job trends data aren't meaningful, and in their magnitudo they are drowned in noise. I had written some counter-arguments about those numbers, but I see that in the meantime Charles has given some insights about that, so I won't repeat more or less the same things. For what concerns the traffic in forums, I'd like to point out that we should count our many localized mailing lists, those about JME and the Platform, things that JDeveloper hasn't.
But I wouldn't discuss in this perspective, nothing meaningful comes out those numbers. I'd instead say that in any case merging two communities makes a larger and more vibrant community. Oracle would do a very smart thing, for both its own interest and the community interest, if it merged the two products: NetBeans would provide an excellent platform, excellent general-purpose IDE capabilities, while JDeveloper of course has got excellent support for Oracle products that could be integrated on top of NetBeans (both IDEs are based on Swing, thanks God, and their UI parts could be mixed in some way). Given that NetBeans supports branding, a JDeveloper merged on top of the NetBeans Platform and IDE could even keep its own identity/brand, exactly as the most recent JBuilder is based on the Eclipse Platform. Oracle could even make money out of the NetBeans Platform and have an additional product in its portfolio, as it is used by a large number of customers.