v1.0.1

Fixes

• Fixed Docker image publishing in CI.

v1.0.0

Features

• Captures inserts, updates, deletes and truncates from PostgreSQL 15+ with no database extensions required.
• Works with partitioned tables and TimescaleDB hypertables.
• Sinks:
  • Kafka, with TLS, SASL, partition keys and headers
  • NATS Core and NATS JetStream
  • RabbitMQ, with delivery confirmation and detection of unroutable messages
  • RabbitMQ Streams
  • HTTP webhooks, with a configurable URL, method, headers and body
• Ships as a single lightweight binary that starts with one command.
• Official multi-architecture Docker image, signed and published with a software bill of materials.
• Configure it with a YAML file, inline config, command-line flags or environment variables.
• Sets up replication on start, and a cleanup mode removes everything it created.
• Configures tables for full change capture automatically. You can turn this off for any table, and Hump warns about tables that are missing the setup.
• Events use a Debezium-compatible format by default.
• Every topic, subject, routing key, header and payload can be templated, with a rich function library for strings, case conversion, conditions and JSON.
• Templates can tell a value the database didn't send apart from a real empty value.
• Reusable sink presets, several tables under one config entry, and a filter that picks which change types each sink receives.
• Your application can attach context, such as trace IDs, to a transaction, and Hump adds it to every event from that transaction.
• At-least-once delivery: a change counts as processed only after every sink has delivered it.
• Each sink has its own retry policy (exponential or linear backoff) and failure policy (stop or skip).
• Snapshots load existing rows into the same sinks:
  • consistent point-in-time data
  • up to 10 tables in parallel
  • an optional row filter per table
  • row estimates before the run starts
  • live changes are held during the run and delivered once it finishes
  • start, stop and watch progress per table and per sink
• Built-in web console:
  • dashboard, flows, statistics and configuration view
  • live event stream with a sink filter you can share as a link
  • snapshot page with a filter editor that has syntax highlighting, column autocompletion and suggested filters
  • light and dark themes
• Prometheus metrics, including replication lag and processing time, plus health probes.