Skip to content

Testing and conformance

GSP separates protocol validity, reference behavior, backend capability claims, and visual review. A rendered example alone is not conformance evidence.

Validation layers

Layer Purpose
Unit and integration tests Validate Python behavior and backend adapters.
Versioned fixtures Preserve protocol records and expected outcomes.
Conformance replay Compare backend results and structured diagnostics.
Visual QA Produce review artifacts without silently promoting capabilities.
API review examples Present small, readable scenes for human comparison.

Run the portable baseline:

uv run pytest
uv run mypy src
uv run ruff check .
uv run mkdocs build --strict

Run the conformance report without requiring a local Datoviz installation:

uv run python tools/conformance_debug_report.py

Normative and operational references