BabelQueue for Go

Polyglot Queues, Simplified. Read and write the canonical BabelQueue message envelope from Go — so your Go services exchange messages with Laravel, Symfony, Python, .NET and Node over one strict JSON format, on the broker you already run.

This is the framework-agnostic Go core: the wire-envelope codec, contracts and dead-letter helpers — zero dependencies (standard library only). The full standard is documented at babelqueue.com.

Installation

go get github.com/babelqueue/babelqueue-go

Requires Go >=1.21.

Usage

package main

import (
	"fmt"

	babelqueue "github.com/babelqueue/babelqueue-go"
)

func main() {
	// Produce — build the canonical envelope and publish the JSON to your broker.
	env, err := babelqueue.Make(
		"urn:babel:orders:created",
		map[string]any{"order_id": 1042},
		babelqueue.WithQueue("orders"),
	)
	if err != nil {
		panic(err)
	}
	body, _ := env.Encode() // []byte of compact UTF-8 JSON
	// redisClient.RPush(ctx, "queues:orders", body)
	//   /  ch.PublishWithContext(ctx, "", "orders", false, false, amqp.Publishing{Body: body})

	// Consume — decode a message produced by ANY BabelQueue SDK.
	in, err := babelqueue.Decode(body)
	if err != nil || !in.Accepts() {
		return // malformed or unsupported — dead-letter / drop
	}
	switch in.URN() {
	case "urn:babel:orders:created":
		fmt.Println(in.Data["order_id"], in.TraceID) // 1042, the cross-service trace id
	}
}

The envelope is identical to every other SDK's:

{
  "job": "urn:babel:orders:created",
  "trace_id": "…",
  "data": { "order_id": 1042 },
  "meta": { "id": "…", "queue": "orders", "lang": "go", "schema_version": 1, "created_at": 1749132727000 },
  "attempts": 0
}

Encode disables HTML escaping and emits compact JSON — the same canonical wire form as the PHP and Python cores (the envelope frame is byte-identical). Key order inside your data map follows encoding/json (sorted), where PHP/Python keep insertion order — JSON object key order is insignificant, so consumers read them identically. JSON numbers decode into Data as float64 (encoding/json's default for any).

Trace continuation

Make mints a fresh trace_id unless you pass one — propagate an inbound trace across a hop with WithTraceID:

next, _ := babelqueue.Make(
	"urn:babel:shipping:requested",
	map[string]any{"order_id": 1042},
	babelqueue.WithTraceID(in.TraceID),
)

Dead-letter

dl := babelqueue.Annotate(env, "failed", "orders", 3, err) // additive dead_letter block
body, _ := dl.Encode()
// publish body to the "orders.dlq" queue

Annotate returns a copy — the original envelope is preserved unchanged inside the dead-lettered message, so any-language consumers can still read it.

Runtime — produce & consume

The core is just the codec. An optional, still-zero-dependency runtime (App) ties it to a broker through a small Transport interface, with URN routing, attempts-based retry → dead-letter, and unknown-URN strategies:

import babelqueue "github.com/babelqueue/babelqueue-go"

app := babelqueue.NewApp(transport, // a Transport (see below)
    babelqueue.WithDefaultQueue("orders"),
    babelqueue.WithMaxAttempts(3),
    babelqueue.WithDeadLetter(true),
)

app.Handle("urn:babel:orders:created", func(ctx context.Context, env babelqueue.Envelope) error {
    // ... handle env.Data; return an error to retry / dead-letter
    return nil
})

app.Publish(ctx, "urn:babel:orders:created", map[string]any{"order_id": 1042})
app.Consume(ctx) // blocks; routes by URN until ctx is cancelled

InMemoryTransport (in the core) backs tests and local runs with zero deps. Broker drivers live in separate modules, so the core itself never pulls a dependency:

go get github.com/babelqueue/babelqueue-go/redis   # Redis  (go-redis)
go get github.com/babelqueue/babelqueue-go/amqp    # RabbitMQ (amqp091-go)

import (
    babelqueue "github.com/babelqueue/babelqueue-go"
    "github.com/babelqueue/babelqueue-go/redis"
)

tr, _ := redis.New("redis://localhost:6379/0")        // reliable BLMOVE + processing list
app := babelqueue.NewApp(tr, babelqueue.WithDefaultQueue("orders"))

The RabbitMQ transport (amqp.New("amqp://…")) publishes to a durable queue with persistent delivery and the contract AMQP properties (type=URN, correlation_id=trace_id, x-schema-version/x-source-lang/x-attempts), and consumes with basic.get + manual ack. Implement babelqueue.Transport yourself to back the runtime with any other broker.

Idempotent consumption (dedupe on meta.id)

BabelQueue is at-least-once, so handlers should dedupe on the envelope's meta.id (ADR-0022). The idempotency subpackage (in the core, zero-dep) makes that a one-line wrap: a message whose id was already processed is skipped instead of re-run.

import "github.com/babelqueue/babelqueue-go/idempotency"

store := idempotency.NewInMemoryStore() // process-local reference store
app.Handle("urn:babel:orders:created", idempotency.Wrap(store, func(ctx context.Context, env babelqueue.Envelope) error {
    // runs at most once per meta.id; a thrown error leaves the id unmarked so retry/DLQ still apply
    return nil
}))

InMemoryStore is for tests / single-process consumers. A fleet of consumers needs a shared, persistent Store — two backends ship as separate submodules (so the core stays zero-dependency; a driver is pulled in only if you import them):

go get github.com/babelqueue/babelqueue-go/idempotency-redis      # Redis    (go-redis)
go get github.com/babelqueue/babelqueue-go/idempotency-postgres   # Postgres (database/sql + pgx)

import (
    "github.com/babelqueue/babelqueue-go/idempotency"
    redisstore "github.com/babelqueue/babelqueue-go/idempotency-redis"
)

store, _ := redisstore.New("redis://localhost:6379/0", redisstore.WithTTL(24*time.Hour))
defer store.Close()
app.Handle("urn:babel:orders:created", idempotency.Wrap(store, handler))

Both persistent backends implement the same idempotency.Store interface as the in-memory reference, with an atomic claim so concurrent consumers serialize on meta.id — Redis via SET key val NX PX <ttl>, Postgres via INSERT … ON CONFLICT … RETURNING on the primary key (DDL via schema.sql / Store.Migrate). A duplicate delivery is detected as already-seen; an optional WithTTL expires ids so they may be re-processed once the window lapses. See each submodule's README for details.

Transactional outbox (atomic write + relayed publish)

A plain producer makes a dual write — commit the business row and publish to the broker — two independent systems that disagree on a crash. The outbox subpackage (in the core, zero-dep) removes it (ADR-0029): the message is persisted into the same database, in the same transaction, as the business data, so it commits or rolls back atomically with it; a separate relay publishes the durable rows afterwards. It is the producer-side counterpart to consumer-side idempotency above.

The caller owns the transaction boundary — Outbox.Write does not begin or commit anything; it just encodes the envelope (frozen codec, bytes unchanged) and hands it to a Store the caller has bound to their open transaction:

import "github.com/babelqueue/babelqueue-go/outbox"

ob := outbox.New(store) // store is YOUR DB-bound outbox.Store (tx-scoped)

// inside the transaction you already opened around the business write:
insertOrder(ctx, tx, order)                  // the business write
env, _ := babelqueue.Make("urn:babel:orders:created", data, babelqueue.WithQueue("orders"))
ob.Write(env)                                // same tx, via a tx-bound Store
tx.Commit()                                  // both, or neither

The outbox.Store interface is the only thing you implement — the core pulls in no DB driver (GR-7). It stores the Encoded envelope verbatim and the relay publishes those exact bytes, so trace_id is preserved end-to-end (GR-4) and every SDK stays byte-compatible (GR-5):

type Store interface {
    Save(encoded []byte, queue string) (id string, err error)
    FetchUnpublished(limit int) ([]outbox.Record, error) // SHOULD claim rows (FOR UPDATE SKIP LOCKED)
    MarkPublished(ids []string) error
    MarkFailed(id, reason string) error
}

A separate relay drains the committed rows through the same Transport seam, on a worker loop or a cron:

relay := outbox.NewRelay(transport, store, outbox.Options{}) // zero Options = sane defaults

relay.Flush(ctx)      // publish one batch; mark each row published AFTER publish returns
relay.Drain(ctx, 0)   // loop Flush until a pass makes no progress (safety ceiling)

A row is marked published only after the transport accepts it — a crash in between re-publishes it next pass (at-least-once; the consumer dedupes on meta.id). A publish error is caught, the row is marked failed (with a bounded linear backoff) and left pending, so one poison row never blocks the batch. outbox.InMemoryStore is a process-local reference for tests / single-process demos (no real transaction — use a DB-backed adapter in production).

GDPR field encryption (encrypt PII inside data)

A registry can declare which data fields carry personal data with the x-gdpr-sensitive schema keyword, and bqschema gdpr audits/masks them (ADR-0030) — but that is governance. The gdpr subpackage (in the core, zero-dep) is the runtime half: a producer encrypts each marked field before publish, a consumer decrypts it after decode, so PII never sits in cleartext on the broker.

The envelope stays frozen (GR-1): Protect rewrites only the values inside data — a sensitive field's value becomes a ciphertext string — it never adds, renames or retypes an envelope field, meta.schema_version stays 1, and trace_id is untouched (GR-4). data remains pure JSON (GR-3), so any SDK can carry the envelope even without the key (it just can't read the protected fields).

The crypto is a Cipher interface the caller provides (KMS / Vault / HSM / tokenisation) — keeping the core dependency-free (GR-7). A stdlib-only AESGCMCipher (AES-256-GCM, random nonce, base64) ships as the reference:

import (
    babelqueue "github.com/babelqueue/babelqueue-go"
    "github.com/babelqueue/babelqueue-go/gdpr"
    "github.com/babelqueue/babelqueue-go/schema"
)

cipher, _ := gdpr.NewAESGCMCipher(key32) // your 32-byte key — or bind a KMS-backed Cipher
provider, _ := schema.NewDirProvider("registry.json")

// Producer — encrypt the marked fields after building data, before publishing.
env, _ := babelqueue.Make("urn:babel:users:registered", data, babelqueue.WithQueue("users"))
if sch, ok, _ := provider.Schema(env.URN()); ok {
    _ = schema.Check(provider, env.URN(), env.Data) // validate CLEARTEXT first (see note)
    _ = gdpr.Protect(env.Data, sch, cipher)         // email, profile.full_name, addresses[].line → ciphertext
}
body, _ := env.Encode()                             // ciphertext rides inside data; frame unchanged

// Consumer — decrypt after Decode, before the handler reads data.
in, _ := babelqueue.Decode(body)
if sch, ok, _ := provider.Schema(in.URN()); ok {
    _ = gdpr.Unprotect(in.Data, sch, cipher)        // restores the original values byte-for-byte
}

Protect/Unprotect are standalone helpers — strictly opt-in, no behaviour change when unused. The sensitive paths come from the same per-URN schema the validation path already loads, including nested objects (profile.full_name) and array items (addresses[].line); an absent field is skipped, not an error.

Validate cleartext, not ciphertext. A schema that constrains a sensitive field (minLength, enum, type:"integer", …) would reject the ciphertext string. Run schema.Check before Protect on the producer and after Unprotect on the consumer — or register schema.Wrap so it runs after your Unprotect step. A wrong-key Unprotect returns gdpr.ErrDecrypt, so the message takes the retry / dead-letter path rather than being handled blind.

This is the reference implementation the other SDKs mirror: the same Cipher seam, the same encrypt-values-in-place contract, the same frozen envelope.

What this core is

It enforces the contract: the envelope shape, URN identity, trace propagation, schema-version gating and the dead-letter block — with zero dependencies. The optional App runtime and InMemoryTransport are zero-dep too; only the Redis and RabbitMQ transport modules pull a broker driver, and only if you import them.

Unknown-URN strategy constants (StrategyFail, StrategyDelete, StrategyRelease, StrategyDeadLetter) drive the runtime's unknown-URN handling.

Conformance

This core passes the shared cross-SDK conformance suite (vendored under testdata/conformance/) — the same fixtures every BabelQueue SDK must satisfy, so a Go producer and, say, a Laravel consumer agree byte-for-byte.

go test ./...

License

MIT © Muhammet Şafak