My personal attitude is that if you are planning on having a large project, one of the following statements is true:

1. You are ready for your probable failure.
2. You have specific reasoning for why what is true for the rest of the industry is not true for you.
3. You are ignorant. (The most likely option.)
4. You are an idiot. (The second most likely option.)

This may sound extreme. I don't believe it is. As I pointed out at RE (tilly) 3 (disaster): Java vs. Perl from the CB, the statistics for failure in software development are so bad it is simply absurd. And the larger the project is, the worse its odds get.

Now this doesn't mean that there are not occasionally large projects that need to be done, despite the horrible odds. But it does strongly suggest that if you can avoid things that lead to large projects, you should. And while Java has features which are intended to help in large projects, it also takes a lot more Java to do the same thing as a little Perl...

UPDATE
In case it wasn't obvious, this is a response to gloom's argument that we "NEED" Java for its features which are meant to make large projects easier. For an alternate response to the same point, see Java Is Untyped. As for the feature, I think that asserts are a very useful strategy. Anything that makes them easier for people to use is good.