[michaelper22]

Here's one of my current use cases: We have a folder tree table in our (formerly nested sets, then closure table, now materialized path), and we cache a materialized version of the path with the folder names. There are some places in the app where you can look up a node by its full path, and this column should conceptually be unique. Obviously an unindexed VARCHAR(2048) is bad to filter on, but MySQL indexes can only cover the first 767 bytes (we used utf8mb4). So we have another column, PathHash CHAR(40) AS (SHA1(path)) VIRTUAL, and then a unique index on that.