Skip to Content
Introduction0.8 Feature Guide

KalamDB 0.8 Feature Guide

KalamDB 0.8 brings PostgreSQL SQL and client connections to the same execution path as HTTP SQL, alongside realtime subscriptions, private tables, shared-table RLS, and background work. This index maps the release capabilities to practical guides.

KalamDB 0.8 feature coverage

CapabilityGuide
Install, local development, generated clients, migrations and deploymentCLI workflow, migrations, functions
PostgreSQL grammar across HTTP and PGWire; prepared statements and introspectionPostgreSQL tools
Schemas, quoted identifiers, search_path, SET / RESET / SHOW and custom session settingsSchemas and settings
USER, SHARED and STREAM tables; optional primary key and hidden row identityTable types, table DDL
Scalar types, arrays, NUMERIC, JSON / JSONB, timestamps, enums, composites and domainsData types, named types and domains
PostgreSQL-style generated IDs and serial-counter helpersSequences and IDs
INSERT / UPDATE / DELETE, RETURNING, upsert, joins, CTEs and query planningQueries, EXPLAIN
Transactions, savepoints, failed-block recovery and isolation boundariesTransactions
Scalar indexes, numeric range seeks, backfill and query performanceIndexes
SQL / PLpgSQL / JavaScript routines, overloads, argument modes and permissionsFunctions and procedures
Topic triggers, scheduled routines, durable consumers, ACK and retriesServer functions, topics
Live SQL, initial state, reconnect and schema-change behaviorLive queries, reconnection
JSON text and KDOM binary frames, generated routine invocationWebSocket protocol, SDK routine calls
JWT / OIDC, normal users, roles and shared-table policiesAuthentication, external provider, RLS
TypeScript / React, Rust, Dart / offline Flutter and Python SDKsSDK guide, Python
FILE uploads and downloads; vectors and embeddingsFiles, vector search
Hot and cold storage, object storage, flush and retentionTiered storage, storage commands
Backup, restore, clustering, monitoring and production configurationBackup, cluster commands, monitoring
Admin UI workflows and runnable application examplesAdmin UI, examples

PostgreSQL compatibility boundaries

The wire listener supports simple queries and extended Parse / Describe / Bind / Execute / Sync / Close, including parameter and row descriptions, text and binary binds, and NULL values. PostgreSQL catalogs expose schema, column, type, routine and index metadata, plus compatibility views for constraints and sequences. Catalog presence alone does not establish enforcement or writable sequence support. Connections retain separate transactions and session settings.

KalamDB is its own database engine. Wire COPY IN / COPY OUT and out-of-band CancelRequest are not supported. Use insert batches, statement timeouts and connection lifecycle controls. SERIALIZABLE and READ UNCOMMITTED isolation levels return 0A000. Standalone CREATE SEQUENCE and nextval are unsupported in the tested server; use supported serial-counter helpers. Table CHECK, secondary UNIQUE, and foreign-key declarations are accepted without enforcing their constraints in that binary. Domain checks are enforced. Range, multirange and shell CREATE TYPE forms are explicitly rejected. A parser accepting PostgreSQL syntax does not mean every PostgreSQL extension or tool works unchanged.

KalamDB-specific features include table TYPE options, FILE, EMBEDDING(N), JavaScript / TypeScript server routines, and SUBSCRIBE TO live queries. Database branching and KalamDB Sites are not released 0.8 features. KoldStore remains a separate experimental product.

Upgrade and release details

Read Upgrade to 0.8 before using an existing data directory. Server, CLI, SDKs and generated clients must use the same 0.8 cohort.

The 0.8 release notes describe the release changes. The compatibility matrix distinguishes implemented behavior, test evidence and remaining gaps. Performance figures live in Benchmarks and Comparisons, with their measured versions and hardware.