# UUIDv7
[![CI](https://github.com/ryanwinchester/uuidv7/actions/workflows/ci.yml/badge.svg)](https://github.com/ryanwinchester/uuidv7/actions/workflows/ci.yml)
[![Hex.pm](https://img.shields.io/hexpm/v/uuid_v7)](https://hex.pm/packages/uuid_v7)
[![Hex.pm](https://img.shields.io/hexpm/dt/uuid_v7)](https://hex.pm/packages/uuid_v7)
[![Hex.pm](https://img.shields.io/hexpm/l/uuid_v7)](https://github.com/ryanwinchester/uuidv7/blob/main/LICENSE)
UUIDv7 for Elixir and (optionally) Ecto, using an 18-bit randomly-seeded counter.
Uses suggestions described in **[Section 6.2](https://www.ietf.org/archive/id/draft-ietf-uuidrev-rfc4122bis-14.html#name-monotonicity-and-counters)** from [this IETF Draft](https://www.ietf.org/archive/id/draft-ietf-uuidrev-rfc4122bis-14.html)
to add additional sort precision to a version 7 UUID.
## When should I use this package?
- You want sequential, time-based, ordered IDs (per-node).
- You are willing to trade a small amount of raw performance for these
guarantees. You are taking a hit for the counter with rollover protection,
and backwards time-leap protection.
NOTE: In this library, sequential UUIDs and ordering are more important than time precision and performance.
We take a slight hit in both of those areas to ensure that the UUIDs are in order. For example, in the case of a
backwards time leap, we continue with the previously used timestamp, and in the case of rollover, we increment
the timestamp by one to ensure that the ordering is maintained.
## When should I not use this package?
- You don't care about sort/order precision beyond milliseconds.
There are other UUID packages, that only have millisecond precision, for example:
- [martinthenth/uuidv7](https://github.com/martinthenth/uuidv7)
- [bitwalker/uniq](https://github.com/bitwalker/uniq)
## Installation
The package can be installed by adding `uuid_v7` to your list of dependencies in `mix.exs`:
```elixir
def deps do
[
{:uuid_v7, "~> 0.5.0"}
]
end
```
## Usage
```elixir
iex> UUIDv7.generate()
"018e90d8-06e8-7f9f-bfd7-6730ba98a51b"
iex> UUIDv7.bingenerate()
<<1, 142, 144, 216, 124, 16, 127, 196, 158, 92, 92, 74, 83, 46, 116, 173>>
```
## Usage with Ecto
Use this the same way you would use `Ecto.UUID`. For example:
```elixir
defmodule MyApp.Blog.Post do
use Ecto.Schema
@primary_key {:id, UUIDv7.Type, autogenerate: true}
@foreign_key_type UUIDv7.Type
schema "blog_posts" do
field :text, :string
# etc.
end
end
```
To use UUIDs for everything in your migrations, it's easiest to just add that as the
default type in your `config.exs`. e.g.:
```elixir
config :app, App.Repo,
migration_primary_key: [type: :binary_id],
migration_foreign_key: [type: :binary_id]
```
## Benchmarks
Run benchmarks with
```
MIX_ENV=bench mix run bench/filename.exs
```
Where `filename.exs` is the name of one of the benchmark files in the `bench` directory.
### Compared to `Uniq.UUID`
(which has no counter or time-leap protection. millisecond precision.)
#### String:
```
Name ips average deviation median 99th %
uniq v7 string 2.23 M 448.71 ns ±3082.24% 417 ns 583 ns
uuid_v7 string 2.08 M 480.89 ns ±3868.08% 417 ns 625 ns
Comparison:
uniq v7 string 2.23 M
uuid_v7 string 2.08 M - 1.07x slower +32.18 ns
```
#### Raw (binary):
```
Name ips average deviation median 99th %
uniq v7 raw 3.35 M 298.15 ns ±7140.23% 250 ns 375 ns
uuid_v7 raw 2.71 M 368.53 ns ±4920.92% 333 ns 459 ns
Comparison:
uniq v7 raw 3.35 M
uuid_v7 raw 2.71 M - 1.24x slower +70.37 ns
```