generic-ml-cache-bootstrap 0.29.0

Composition root and plugin discovery for generic-ml-cache: discovers gmlcache.adapters plugins (with load-time whitelisting and provenance trust) and wires the application.

generic-ml-cache-bootstrap

The composition root and plugin discovery for gmlcache

The startup layer that assembles the application. It discovers the gmlcache.adapters plugins (whitelisting them at load time and judging which distributions are trusted), then wires the core use cases to the concrete adapters and hands the drivers a ready application API.

It is the one place that freely wires both core and adapters — the composition root in hexagonal terms. The drivers (cli, daemon) get their wired application from bootstrap and reach the domain only through the inbound ports of that ApplicationApi bundle; the DB-backed adapters are grouped behind an injectable PersistenceBackend, so no driver ever touches a raw connection. A driver keeps only a few deliberate direct edges to adapters, each a driver-policy choice rather than domain wiring: the CLI's own encrypt/decrypt/rotate command helpers (the crypto adapters) and its demo stub adapter, and each driver selecting its own diagnostics (logging) adapter. An adapter never imports bootstrap (that would be a leaf calling the composition root). These edges are frozen by import-linter contracts.

Install

You normally get it transitively (pip install generic-ml-cache-cli pulls it in). Installed on its own:

pip install generic-ml-cache-bootstrap

License

Apache-2.0 — see LICENSE and NOTICE.