Skip to content

Latest commit

 

History

History
26 lines (24 loc) · 1.54 KB

File metadata and controls

26 lines (24 loc) · 1.54 KB

spawn-ts documentation

  • Concepts — the cost-safety model for newcomers: TTL vs idle vs cost limit, why TTL always terminates, the absolute-deadline invariant.
  • Architecture — layers, the provider seam, design principles.
  • Lifecycle engine — the priority order and invariants ported from spore.host spored.
  • Parameter sweeps — fan a parameter grid out into many instances; the spec format, the spawn:sweep-* tag contract, and the shared fan-out engine.
  • Batch job queues — launch a DAG of jobs as dependencies complete and capacity allows; the config format, dependency gating, retries, and the on-failure policy.
  • Execution shapes — single node, job array, MPI. Job arrays with --min-viable and sparse-index accounting; why MPI is tags only.
  • API reference — the SpawnClient public API. The generated TypeDoc reference is published alongside the demo.
  • Data movement and plugins — browser-native Globus Transfer (no local machine required), and the plugin split: 7 of 12 declarable at launch, 12 of 12 detectable via the spore:plugin:* tag.
  • Integration with truffle-ts — how the launcher/lifecycle (spawn-ts) and instance-discovery (truffle-ts) tools compose, and the tag-emit-vs-execution boundary.

For a quick start and backend setup, see the top-level README.