When I first started web development (mid/late 90's), Postgres was the very conservative open source choice. MySQL was fast, good enough, and broke a few big rules.
That hasn't been the case for a while, but with the recent speed improvements in the past few years, prompted by NoSQL, and MySQL's troubles with Oracle, Postgres has become the Open Source SQL frontrunner, and everybody's taking it in really interesting directions.
I'm hoping the JSON & plv8 support gets much more in the box, and that the replication setup becomes more baked in as well.
Right now, I'm migrating from MS-SQL to a MongoDB replica set for most of our core data. Mainly because the failover works better than most, and the licensing costs for even MS-SQL with replication and failover are budget blowing. Most of the PostgreSQL options for replication are really bolted on with some serious drawbacks, and automatic failover to a new master is another issue.
I've always liked PostgreSQL (and Firebird for what that's worth)... I do hope that some of these features become more of a checkbox item during installation, and less of a bolted on, have to compile and dive into the deep to get them, and even then only have it half baked.
I also started development in the 90s. Postgres became an interesting choice in the mid-2000's once (someone can correct me) the organization responsible with stewarding it did a re-org and opened up more to outsiders.
If I remember correctly, Postgresql had stored procedures way before Mysql AND you could create them with Python and several other languages.
