We Chose NanoIDs for PlanetScale's API

The post dismisses UUIDs because their default formatting uses hyphens and hex encoding (base16), but as other comments mention you can get UUIDs to be much shorter by encoding them in a different base. If you want to guarantee URL-safety and selection via double-click, use base62 (26 lower-case, 26 upper-case, 10 digits).

  uuid  : c5831108-6f95-442a-b2b0-3c62dc0abdd1
  base62: 60hNvkHrZykznfLaESuDBp

This format also lets you attach a prefix with '_' so you can keep track of the semantic type of each ID. This pattern is widely used by the Stripe API, and I've grown quite fond of it (so much easier to grep).

  url: api.planetscale.com/v1/deploy-requests/deploy_60hNvkHrZykznfLaESuDBp

Additionally, within an in-memory value or a database, you probably want to store the decoded form of any ID. The value "deploy_60hNvkHrZykznfLaESuDBp" is 29 bytes long, but could be represented by `type DeployID byte[16]` for ~50% space savings and stronger type validation.

So UUIDs were the perfect solution except that, if a user double-click a url with a UUID in it, it won’t select all of it.

Can’t you just… remove the hyphens? UUIDs are just arrays of bytes, the hyphens are not part of the data.

Instead, they use a less popular, less supported, more collision-prone ID format.

We use crypto.strong_rand_bytes in elixir and Base.encode32, strip padding.

It works like a charm. Length decided by setting number of random bytes.