[adgjlsfhk1]

specifically, they are one of the coolest features in postgres. Have you ever wanted to pretend that a CSV (or your filesystem or google) or basically anything else was a SQL database? With a couple hundred lines of code, you can do this. The performance won't be great, but it's ridiculously flexible.

[ComputerGuru]

I have a two-liner shell script that turns a csv into a virtual SQLite db (./to_sql.sh foo.csv) and lets me use the full power of SQLite. Usually that’s enough for my purposes, but I do use pgsql‘s copy from csv to generate tables representing foreign data imported from a csv (say census info or NAICS occupation codes).