Rift runs as a self-contained, statically compiled binary or containerized job. No background daemons, no database write access, and no permanent application tier required.
Select commands below to inspect real outputs, exit codes, and JSON payloads.
Compare two datasets exactly and surface discrepancies without full-table diff dumps.
-s, --schema <FILE>: Path to canonical schema JSON file.-k, --key <COLUMNS>: Primary key column name(s). Comma-separate for composite keys (e.g. --key org_id,user_id).--unkeyed: Reconcile datasets as unkeyed multisets (using hash multisets).--engine <MODE>: Execution engine (in-memory or external). Default: in-memory.--memory-limit-mb <MB>: Sorter buffer allocation budget for external sort (default: 256 MB).-o, --output <FORMAT>: Output format (text or json).--non-interactive: Disable interactive prompts (recommended for CI/CD and AI agents).Automatically discover candidate RTL explanations from observed dataset differences.
When datasets diverge, rift explain initiates a multi-stage Parameterized Graph Discrepancy Search. It discovers candidate invariants across divergent attributes (e.g., fee subtractions, percentage discounts, status transitions) and verifies candidate rules via an authoritative verification pass.
Rift requires an explicit schema JSON declaring fields and canonical primitive types.
{
"fields": [
{
"name": "id",
"type": "string"
},
{
"name": "amount",
"type": "decimal",
"precision": 18,
"scale": 2
},
{
"name": "account_id",
"type": "int64"
},
{
"name": "created_at",
"type": "timestamp"
},
{
"name": "is_settled",
"type": "boolean"
}
],
"primary_key": ["id"]
}| Exit Code | Constant Name | Semantics & Behavior |
|---|---|---|
| 0 | EXIT_OK | Datasets match exactly (zero discrepancies detected) or 100% verified explanation. |
| 1 | EXIT_DISCREPANCIES | One or more discrepancies found (INSERTED, DELETED, MODIFIED) or partial explanation. |
| 2 | EXIT_ERROR | Operational or user input error occurred (file not found, schema mismatch, invalid syntax). |
Rift is proprietary commercial software developed and licensed by Eagle Eye Research and Deployment Labs Private Limited. Binaries and registry access are distributed privately under enterprise evaluation or commercial licensing agreements. Prior Apache-2.0 grants on historical releases (v0.1.1 and earlier) remain valid for those historical artifacts.