slid

slid is a simple, 12-byte (96-bit) lexicographically-sortable identifier encoder for Go that embeds a nanosecond-resolution timestamp with a 32-bit atomically-incremented counter value. In other words, it generates IDs. Which are small. And probably unique (usually 2.14e22 unique identifiers per millsecond). And maybe usable.

Lexicographically-ordered IDs can be quite useful for key/value stores like LMDB and BoltDB in situations when you'd like for values to be time-sorted. The identifiers generated by slid sacrifice human readability for performance and identifier size versus, for example, RFC 3339-encoded timestamps.

NOTE: Since a SLID is generated via a counter value, it's best used only in long-running processes, and only when there's no vulnerability in a user being able to potentially jump an identifier to the next value in the sequence. Otherwise, prefer some sort of UUID.

Getting Started

This package uses Go modules. If you've installed Go, import the package:

import "github.com/boatilus/slid"

You can then run go mod tidy to grab the package.

From there, slide into generating some slids ⛷️

id := slid.New()                    // constructs a new SLID struct with the current time
encodedID := id.Encode()            // returns a []byte
decodedID := slid.Decode(encodedID) // returns a SLID struct

Benchmarks

BenchmarkNew-4             	50000000	        30.2 ns/op
BenchmarkEncode-4          	30000000	        34.7 ns/op
BenchmarkNewThenEncode-4   	30000000	        52.3 ns/op
BenchmarkDecode-4          	100000000	        10.9 ns/op

(Run on a 2.7 GHz Haswell Core i5)

Running the tests

go test -v

License

This project is dedicated to the public domain via Creative Commons CC0 - see LICENSE.md for details.

Acknowledgments

• Inspired by Alizain Feerasta's ULID.