ash_prefixed_id

# Changelog

## 0.2.0

### Fixed

- Casting/decoding a malformed prefixed ID whose slug contains characters
  outside the base58 alphabet (`0`, `O`, `I`, `l`, punctuation, …) no longer
  crashes with `ArithmeticError`/`ArgumentError` from erl_base58 — every
  malformed input now returns `{:error, :invalid_prefixed_id}` (or raises
  `ArgumentError` from the bang variants). This is the exact class a public
  URL parameter delivers to a cast path.

### Added

- `to_uuid/1` and `to_uuid_string/1` — non-bang variants returning
  `{:ok, value} | {:error, :invalid_prefixed_id}`, for validating external
  input (HTTP params, etc.) at system boundaries.
- `to_prefixed_id/2` now also accepts a resource module as the second argument,
  resolving the prefix from the resource's `prefixed_id` DSL (avoids hard-coding
  the prefix string).

### Changed (breaking)

- Removed `decode_object_id/1`. It was a redundantly-named twin of the new
  `to_uuid_string/1`; use `to_uuid_string/1` (tagged tuple) or
  `to_uuid_string!/1` (raising) instead.
- `to_uuid_string!/1` now raises `ArgumentError` for invalid input (previously
  could leak a `MatchError`).
- Conversion failures now use tagged `{:error, :invalid_prefixed_id}` tuples
  instead of a bare `:error`.

## 0.1.1

### Changed

- `PostgresExtension`: drop `LEAKPROOF` from `timestamp_from_uuid_v7/1`.
  LEAKPROOF only matters when a function appears in a query crossing a
  PostgreSQL security barrier (RLS, `security_barrier` views), which Ash
  does not enable by default. Marking a function leakproof requires
  superuser, which blocks every managed Postgres provider (RDS, Cloud
  SQL, Heroku, Supabase, Neon, …). Defaulting to portable migrations is
  worth the lost optimizer hint. Users who actually need it can run
  `ALTER FUNCTION timestamp_from_uuid_v7(uuid) LEAKPROOF;` themselves
  as superuser.

## 0.1.0

Forked from [ash_object_ids](https://github.com/drtheuns/ash_object_ids) and renamed to `AshPrefixedId`.

### Added

- `AnyPrefixedId` type — a universal `:uuid` replacement that accepts any prefixed ID or raw UUID
- `to_uuid!/1` and `to_uuid_string!/1` for converting prefixed IDs to raw UUID formats
- `to_prefixed_id/2` for converting UUIDs back to prefixed form
- `find_resource_for_prefix/2` and `find_resource_for_id/2` for resource lookup by prefix
- `find_duplicate_prefixes/1` and `map_prefixes_to_resources/1` for prefix management
- `PostgresExtension` with `uuid_generate_v7()` for server-side UUIDv7 generation
- `migration_default?` option for automatic PostgreSQL migration defaults
- Automatic `belongs_to` foreign key type inference via `BelongsToAttribute` transformer

### Changed

- Renamed `AshObjectIds` → `AshPrefixedId` throughout
- Moved FK type updates from transformer to persister for reliability
- Improved handling of self-referential foreign keys
- Fixed 16-byte binary handling in `cast_stored`
- Fixed binary-to-string UUID conversion in `cast_input`