Figure 18 — Extension points

Vector PDF

File: fig18_extensibility.pdf

Suggested caption. Extension points. Every part of the benchmark is registered by name. A plugin file registers new components, and the configuration selects them. A new pool generator, assembler, seam joiner or long-series generator becomes part of the comparison and of the relevant hypotheses without changes to the benchmark code.

How it works

• A plugin (my_plugins.py) uses @registry.<kind>.register("name"), or the helpers register_task, register_method, register_fidelity_group, register_test, register_hypothesis.
• Config loads the plugin everywhere: in every notebook, in subprocesses of other environments, and in the hypothesis part (the run configuration is stored in tables/config.json).
• Automatic expansion: pool_suites adds the whole assembly suite on a new pool generator (<prefix>_zseries, <prefix>_rand, …); extra_assemblers adds an assembler on every pool, extra_joiners a seam processing over the assemblers of smooth_paths; long_methods adds a long-series generator as a method and a baseline; disabled switches off any method or component by name. New methods enter the ranking and the research questions automatically (their family is read from the method spec); the confirmatory family covers Diffusion-TS and Chronos.
• Any method can be the subject of the case study (hypotheses.run --case-method … --noise-pair …).
• Source: marathonts/bench/registry.py, EXTENDING.md.