rdt-model-compile

rdt-model-compile is the pipeline orchestrator. It spawns all rdt-model-* module binaries in the correct phase order, captures their JSON result envelopes, and aggregates results.

No generators, templates, or schemas live here — those belong in the module that owns each artifact.

Property	Value
Binary	rdt-model-compile
Pipeline phase	Orchestrator (invokes all phases)
Depends on	All module binaries available on $PATH
Network	None (delegates to child modules)
Threading	Pattern A (async — spawns child processes in parallel within each phase)

Subcommands

Subcommand	Description
run	Execute the full 6-phase pipeline for an entity
run --phase <name>	Execute a single phase only
run --workspace <path>	Run with workspace isolation (modules write to workspace, not repo)
promote --workspace <path>	Copy artifacts from workspace to final repo paths after a successful run
status	Show pipeline status for an entity (not yet implemented)

Execution semantics

• Phases run sequentially: Ingest → Enrich → Prepare → Deploy → Register → Support
• Modules within the same phase run in parallel (futures::join_all)
• If ALL required modules in a phase fail → pipeline stops (status: Error)
• Optional module failures (profile, infer) never block the pipeline
• --dry-run → logs planned invocations, returns synthetic Skipped results
• --json → emits PipelineResult to stdout

Usage

Dry-run the full pipeline

cargo run -p rdt-model-compile -- --target dev --entity waste-tracking --dry-run run

Run a single phase

cargo run -p rdt-model-compile -- --target dev --entity waste-tracking run --phase deploy

Run with workspace isolation

cargo run -p rdt-model-compile -- --target dev --entity waste-tracking \
  run --workspace /tmp/rdt-waste-tracking-test

Modules write to the workspace instead of directly to repo paths. Combine with promote for atomic deployments.

Promote artifacts to repo

cargo run -p rdt-model-compile -- --target dev --entity waste-tracking \
  promote --workspace /tmp/rdt-waste-tracking-test

Copies workspace artifacts to their canonical repo paths. Only promotes files that exist — optional modules that didn’t run are skipped.

JSON output for automation

cargo run -p rdt-model-compile -- --target dev --entity waste-tracking --json run --dry-run

Returns a PipelineResult JSON envelope with per-phase and per-module status.

Global flags

Flag	Short	Env	Description
--target	-t	RDT_TARGET	Required. dev, test, or prod
--entity	-e	—	Entity slug (required)
--workspace	—	—	Enable workspace isolation mode
--run-id	—	RDT_RUN_ID	Pipeline correlation ID (auto-generated if omitted)
--dry-run	-n	—	Skip all side-effects
--json	—	—	Emit structured PipelineResult to stdout
--quiet	-q	—	Suppress informational output

Pipeline phases

#	Phase	Modules	Optional
1	Ingest	pull, profile	profile
2	Enrich	govern, infer	infer
3	Prepare	validate	—
4	Deploy	store, policy, api, mcp, sdk, contract	—
5	Register	register, snowflake-horizon, rdm, gupri, search	—
6	Support	docs, cidb, event	—

Error handling

The orchestrator captures errors from child modules and reports them in the result envelope. A phase-level failure (all required modules fail) halts the pipeline immediately. Individual module failures within a phase produce a Partial status but allow the pipeline to continue if at least one required module succeeds.

Multi-entity orchestration

For running multiple entities in parallel via CI, see:

• Running Multi-Entity Pipelines — practical guide
• Pipeline Configuration — entity list, manifest schema, environment variables