v0.4.0

Retries now cover about 28 hours by default, and a full scope reports its own error code. 0.3.0 announced both changes, and this release follows the 0.x minor rules. The database schema stays at version 5, so no SQL is needed. ADR 0013 records the schedule.

Upgrade

1. Restart producers and workers on 0.4.0. No migration is required.
2. If you set retentionMs below 48 hours without retry.maxAgeMs, set retry.maxAgeMs too. Otherwise the constructor throws INVALID_CONFIG, because retention must cover the age limit.
3. If you check for INVALID_STATE when creating an endpoint in a full scope, check for ENDPOINT_LIMIT. A custom ManagementRepository should throw ENDPOINT_LIMIT at the limit and INVALID_INPUT for a limit outside 0 to 1,000.

Breaking

• Longer default retries. retry.delaysMs defaults to 5 s, 5 min, 30 min, 2 h, 5 h, 10 h, 10 h: eight attempts over about 28 hours, instead of seven over about 43 minutes. retry.maxAgeMs defaults to 48 hours (was 24). Set both to keep the old schedule.
• ENDPOINT_LIMIT. Creating an endpoint in a scope at its maxEndpoints limit throws the new ENDPOINT_LIMIT code instead of INVALID_STATE, so apps can show their own message (#74). Better Auth still answers 409, with code ENDPOINT_LIMIT.

Fixed

• The PostgreSQL repository reports a limit outside 0 to 1,000 as INVALID_INPUT, like the Better Auth repository.

npm install good-webhooks@0.4.0