Skip to content

Bundles

The AssessmentBundleRepository stores algorithm bundles that power the adaptive assessment engine. Bundles are fetched from the Go diagnostics service via gRPC and cached in PostgreSQL.

Model

pub struct AssessmentBundleRow {
    pub id: String,
    pub network_index: i32,
    pub curriculum_index: i32,
    pub data: serde_json::Value,  // AlgorithmBundle JSON
    pub version: i32,
    pub created_at: DateTime<Utc>,
    pub updated_at: DateTime<Utc>,
}

pub struct AssessmentBundleSummaryRow {
    pub id: String,
    pub network_index: i32,
    pub curriculum_index: i32,
    pub version: i32,
    pub created_at: DateTime<Utc>,
    pub updated_at: DateTime<Utc>,
}

Bundles are keyed by (network_index, curriculum_index) — each pair identifies a specific skill graph and curriculum combination.

Repository Methods

Method	Purpose
`get_by_position(network, curriculum)`	Fetch full bundle data
`upsert(network, curriculum, data)`	Insert or update (UPSERT)
`list()`	List all bundles (metadata only, no data blob)
`delete_by_position(network, curriculum)`	Delete bundle
`get_version(network, curriculum)`	Check bundle version without fetching data

UPSERT

pub async fn upsert(
    &self,
    network_index: i32,
    curriculum_index: i32,
    data: serde_json::Value,
) -> Result<()> {
    // INSERT INTO assessment_bundles (id, network_index, curriculum_index, data, version)
    // ON CONFLICT (network_index, curriculum_index)
    // DO UPDATE SET data = $4, version = version + 1, updated_at = now()
}

Each upsert increments the version number, allowing clients to check if their cached bundle is stale.

Data Flow

Go Diagnostics → gRPC → GrpcBundleFetcher → upsert() → PostgreSQL
                                                            ↓
Client ← GraphQL query ← get_by_position() ← ─────────────┘
                                                            ↓
                                            AlgorithmBundle::from_json()
                                                            ↓
                                            AlgorithmEngine (in-memory)

The data column stores the full AlgorithmBundle as JSON — including sparse distance/heuristic matrices and the question pool. See Algorithm for how this data is used.

Migration

Migration	Purpose
`005`	Create `assessment_bundles` table with unique `(network_index, curriculum_index)` constraint

Resources — resource storage
Folders — folder organization
gRPC — how bundles are fetched from diagnostics
Algorithm — how bundles power adaptive question selection