Audit Log

Quark can record an audit trail of every Create, Update, and Delete into a quark_audit table. Turn it on with Client.EnableAuditLog; from then on each write inserts an audit row on the same connection/transaction as the write itself, so the trail is atomic with the data.

err := client.EnableAuditLog(ctx, quark.AuditConfig{
    UserFromContext:   func(ctx context.Context) string { return userID(ctx) },
    TenantFromContext: func(ctx context.Context) string { return tenantID(ctx) },
})

What gets recorded

EnableAuditLog migrates the quark_audit table (idempotent). Its columns:

Column	Meaning
id	Audit row PK.
ts	UTC timestamp of the write.
tenant_id	From AuditConfig.TenantFromContext (empty if unset).
user_id	From AuditConfig.UserFromContext (empty if unset).
table_name	The audited table.
operation	created / updated / deleted.
pk	The affected row's primary key (composite PKs joined with :).
diff	JSON change payload (see below).

The table is created from a model, so its DDL is portable across all six dialects — diff lands in the engine's JSON column (or a text column where the engine has none). You do not write PostgreSQL-specific JSONB/BIGSERIAL DDL by hand.

The diff payload

Operation	diff shape
created	The full inserted row: {"id": 1, "name": "foo", "qty": 3}.
deleted	The full row being deleted.
updated via Update / UpdateFields	The new values: {"name": "bar"} (no prior value — there is no snapshot).
updated via Tracked.Save	A per-column delta: {"name": {"old": "foo", "new": "bar"}}.

To get the {old, new} delta, load through the dirty-tracking API:

tracked, _ := quark.For[Order](ctx, client).Track().Find(id)
tracked.Entity.Status = "shipped"
tracked.Save(ctx) // audit diff = {"status": {"old": "pending", "new": "shipped"}}

JSON number round-tripping

Numbers in the diff come back from the JSON column as float64 when you read them into a map[string]any. Compare them accordingly (or decode into a typed struct).

Atomicity — written with the commit, not after

The audit row is inserted inline on the CRUD connection, so:

• Inside Client.Tx, the audit INSERT joins that transaction. If the transaction commits, the data and its audit row commit together; if it rolls back, both disappear. You never get committed data without its trail, nor a trail for work that was undone.
• Outside a transaction, the audit INSERT is a separate statement immediately after the write. There is a small crash window between the two — wrap your writes in Client.Tx when you need the guarantee.

This is intentionally a stronger guarantee than the Event Bus, whose post-commit emission can be lost on a crash. Losing an event is tolerable; losing an audit record is not.

Filtering which tables are audited

client.EnableAuditLog(ctx, quark.AuditConfig{
    IncludeTables: []string{"orders", "payments"}, // only these
    // or:
    ExcludeTables: []string{"sessions"},           // everything but these
})

ExcludeTables takes precedence over IncludeTables. The quark_audit table is always excluded — auditing the audit log would recurse.

Limitations

• Bulk and WHERE-based methods are not audited: CreateBatch, UpdateBatch, DeleteBatch, DeleteBy, and UpdateMap have no per-row entity to diff. Loop the single-row methods if you need a trail for them.
• Plain Update records new values only — the {old, new} delta requires Tracked.Save, which carries the snapshot.
• No automatic retention/pruning. quark_audit grows unbounded; schedule your own retention job.
• The audit row write uses a raw parameterised INSERT that bypasses the query observer/middleware chain (so it never recurses and never appears in slow-query logs). Identifiers are fixed constants, values are bound — no injection surface.