Execution Pipeline

Summary

The runtime path is centered on RunService in internal/app/run.go. This is the best place to understand the real operational sequence for test, check, and ci.

Runtime sequence

1. Load inputs

Meridian validates global and runtime options, expands config sources, loads env values, and attempts repo-side config loading.

2. Prepare artifacts and engine

It allocates a run directory, reserves runtime ports, and resolves the selected runtime engine.

3. Run semantic validation

It validates the config against the selected Collector target and records stage results and findings.

4. Select the runtime config

It chooses the runtime config source:

• repo-side merged config when available
• Collector-rendered config when required and available

5. Write static artifacts

It runs repo-side validation and graph generation on the runtime config model and stores those results in the pending run result.

6. Patch the runtime config

It injects the Meridian receiver/exporter pair, adjusts selected pipelines, and builds a TestPlan.

7. Attach diff analysis

When diff inputs are present or the command includes diff-aware behavior, it attaches diff results to the run.

8. Execute the runtime harness

The runtime runner starts the Collector, injects synthetic telemetry, waits for capture, persists normalized capture output, evaluates assertions and contracts, and collects logs.

9. Write the bundle

Reporting writes report.json, summary.md, diff and contract artifacts, semantic artifacts, and the latest symlink.

Why this matters

This explicit pipeline is the backbone of both the docs and the codebase. When describing a feature, the docs should make it clear where that feature sits in the sequence.