If you spend some time browsing the documentation you will find some very cool stuff. PostgreSQL looks like a Unix programmer's dream come true. It seems to be very tied-in to Unix networking concepts, and even has a functions for manipulating IP address data stored in columns(
It would be the perfect database engine for managing a large network. It also has extended math features, such as geometric transformation functions (
In fact, with Postgres, a column isn't restricted to single values, but can have arrays of multiple values per column (
So Postgres's value isn't just in how well it acts like a traditional ACID-compliant RDBMS, but also in it's complete feature set, some of which I haven't seen in any other database. I'm planning to make the jump from MySQL to PostgreSQL for my more complex apps.
Not that I regret at all having worked with MySQL in the past. MySQL may be simple, but that is also a strength, especially when putting web-based systems together fast. You get easy portability, replication, and a very concise, straightforward API. And I have never, ever had a problem with data corruption or availability. I'm looking forward to MySQL 4, when they get row-level locking, transactions, and hopefully Foreign Keys worked out. (all are in development now)
They are both good at what they do. MySQL chose to focus on performance and robustness first, and is now starting to add the enterprise level features; PostgreSQL focused on the Serious features first, and is now catching up in the performance department. I'll bet in another year, both will be giving the commercial databases a real run for their money.