Skip to content

contributing.md

Latest commit

 

History

History
63 lines (46 loc) · 3.43 KB

contributing.md

File metadata and controls

63 lines (46 loc) · 3.43 KB
Raw

Contributing & Tests

Running Tests

The project uses pytest with a modular fixture system (tests/conftest.py) that provides shared model runners, port cleanup, and per-test isolation. All tests are collected from tests/, tests/unit/, and any module-scoped test files.

# Run all tests (fast + slow)
python -m pytest -v --tb=short

# Run only fast tests (skips models that download/run LLMs)
python -m pytest -v --tb=short -m "not slow"

# Run only slow integration tests (requires a working GPU/backend or container runtime)
python -m pytest -v --tb=short -m slow

Fixture Summary

Fixture Scope Purpose
mr module Shared ModelArkestra instance — port allocation and runner maps are shared across the module
_cleanup_after_test function (autouse) Stops all models after each test so every method sees a clean slate
_cleanup_ports module (autouse) Safety net that kills any lingering processes on configured ports before/after each module
podman_cleanup function Tracks podman containers/tasks/ports with guaranteed teardown — opt-in per-test fixture

Import Path

The runner and its backends live in the model_arkestra package:

from llm_config_manager.config_manager import ConfigManager    # data layer
from model_arkestra.arkestra import ModelArkestra              # orchestration (recommended)
from model_arkestra.base import BaseModelRunner                # abstract base class
from model_arkestra.process import ProcessModelRunner          # process runner
from model_arkestra.podman import PodmanModelRunner            # podman runner
from model_arkestra.docker import DockerModelRunner            # docker runner
from model_arkestra.container_runner import ContainerModelRunner  # container base class
from model_arkestra.http_client import ModelHttpClient         # lightweight HTTP client
from model_arkestra.langchain_adapter import LangChainModelAdapter  # LangChain LCEL wrapper
from model_arkestra.server import ArkestraServer             # OpenAI v1-compatible API server

# Convenience re-exports from __init__.py:
from model_arkestra import RunnerState, RunnerError, ServerReadyTimeout
from model_arkestra import ModelNotStarted, MaxRestartsExceeded, ModelShutdown

Shared Utilities (common.py)

Functions used by the runner classes to build commands from configuration data:

Function Description
build_model_args(cm, model_name, env_vars=None, override_backend=None) → (list[str], str) Builds command arguments for a model using its backend config — resolves the effective backend, fills ${PORT} and ${CHECKPOINT} placeholders in the backend args template, and concatenates backend args with model-specific args. Returns (arg_list, cmd_str).
_resolve_backend(cm, model, model_name, override_backend) → str Determines which backend ID to use for a model. Resolution order: override_backend → model["backend"] → backends.default. Used internally by build_model_args.

These replace the former ConfigManager.assemble_command() and ConfigManager._resolve_backend_for_model() methods, keeping command assembly logic inside arkestra where it belongs.

Related Documentation