To simplify and standardize data setup in integration and end-to-end tests, we should integrate support for test fixtures—structured, version-controlled YAML (or JSON) files that define initial database state (spaces, tuples, indexes, etc.) for test scenarios.

This will allow test authors to:

• Declare expected dataset declaratively (instead of writing repetitive Insert/Replace boilerplate) and will obsolete (or extend) old way of InsertOnInstances
• Reuse common setups across multiple test suites
• Reset cluster state reliably between tests
• Keep test logic focused on behavior, not data provisioning

Proposed Approach

Introduce a new testfixtures package (or extend test_helpers) that can:

• Load fixture files
• Parse them into typed Tarantool operations
• Optionally create spaces from schema in fixtures
• Apply them to a given *tarantool.Connection or *pool.ConnectionPool
  (with or without space prefix for isolation)
• Optionally clean up (truncate spaces) before or after.

Example

schema:
  users:
    format:
      - {name: id, type: unsigned}
      - {name: name, type: string}
      - {name: role, type: string}
    indexes:
      primary:
        type: TREE
        unique: true
        parts:
          - {field: id, type: unsigned}
      name:
        type: TREE
        unique: false
        parts:
          - {field: name, type: string}
    options:
      if_not_exists: true
  sessions:
    format:
      - {name: id, type: string}
    indexes:
      primary:
        type: TREE
        unique: true
        parts:
          - {field: id, type: string}

spaces:
  users:
    - [1, "alice", "admin"]
    - [2, "bob", "user"]
  sessions:
    - ["sess1", 1, 1700000000]

// Load and apply fixtures
err := testfixtures.Load(conn, "fixtures/basic.yaml")
require.NoError(t, err)

// Now run test logic against pre-seeded data
user, err := getUserByID(conn, 1)
require.NoError(t, err)
assert.Equal(t, "alice", user.Name)