routix 0.0.11

Define routines as structured algorithmic workflows

routix

Routix is a lightweight Python toolkit for designing and executing structured algorithmic workflows.

Key Features

• Subroutine-based execution control: Flexible workflow management via SubroutineController
  • Context-aware logging: Detailed logging with routine context traceability via MethodContextManager
• Structured flow validation: Validate workflow definitions with SubroutineFlowValidator
• Dot-accessible configuration/data objects: Manage hierarchical data and configuration with DynamicDataObject
• Experiment summary and timing: Manage experiment results and timing with ExperimentSummary, ElapsedTimer, etc.
• Extensible runner base classes: Build custom workflow runners (single/multi/concurrent) in src/routix/runner/
• Metric time series management: Collect and store time series data during experiments with MetricTimeSeries and NamedTimeSeriesStore
• Utilities: Tools for saving results/configuration as YAML/JSON and more

Subroutine Flow Data Format

Routix executes workflows defined as structured lists of dictionaries. Each step is clearly specified with method names and parameters.

- method: initialize
- method: repeat
  params:
    n_repeats: 3
    routine_data:
      - method: sample_method
        params:
          value: 42

See subroutine_flow_data.md for details.

Runner Base Classes

• SingleInstanceRunner: Abstract base for running a single problem instance
• MultiInstanceRunner: Abstract base for running multiple instances in sequence
• MultiInstanceConcurrentRunner: Abstract base for running multiple instances concurrently (in parallel)

All runners are designed for subclassing and method overriding to fit your experiment patterns.

Note: InstanceSetRunner is a deprecated name. Please use MultiInstanceRunner instead.

Metric Time Series

• MetricTimeSeries: Manages (timestamp, value, note) time series data
• NamedTimeSeriesStore: Stores and manages multiple named MetricTimeSeries

This enables structured recording of experiment metrics, with export to YAML/JSON supported.

Utilities

• object_to_yaml, object_to_json: Save experiment results and configuration to files
• Additional helpers for experiment management

Testing

Unit tests for all major components are included in the tests/ directory. Run all tests with pytest.

---