The sort order of text in an SQL database is influenced by so-called collations. These usually work behind the scenes, but they offer a number of interesting customization possibilities, while also being a potential source of operational problems and user confusion. In this presentation, I want to start by explaining how modern collations work and what the role of Unicode is. Then we'll discuss what variants and customization options exist, such as language-specific collation and case insensitivity, and how PostgreSQL implements those. Here we can also take a look ahead at relevant development projects for future PostgreSQL versions. Finally, we'll look at how collations affect the management of databases and particularly upgrades. There are some traps here that could cause index corruption or data loss.
