Measuring Higher-Order Strategy Efficacy

[Andrewyx]

Description of the task

The proposed system seeks to formalize existing patterns for backward-compatible simulator interfaces and integrate them with CI/CD, such that rigorous testing/prototyping of higher-level plays can be evaluated definitively.

See the corresponding RFC

Class Diagram

• RuntimeManager should be a singleton class
• RuntimeInstaller should be a delegate of RuntimeManager
• RuntimeLoader should also be a delegate of RuntimeManager
  Reminder that delegates should contain the underlying implementation of the common class RuntimeManager and RuntimeManager should purely call its delegates to fullfill external method calls. Conventionally, public methods in the common class i.e. RuntimeManager share the same name as the methods in its delegates who implement the functionality. E.g. RuntimeManager.fetch_releases() calls -> RuntimeInstaller.fetch_releases().

External Binaries

External Binaries should be stored in the /opt/tbotspython/external_ai directory.

/opt/tbotspython/
├─ external_ai/
│  ├─ ai_config.toml
│  ├─ unix_full_system-0.0.1
│  ├─ unix_full_system-1.10.9
/home/....../bazel-out/k8-fastbuild/bin/software/
├─ unix_full_system

Note

/home/....../bazel-out/k8-fastbuild/bin/software/ is a path to the bazel sandbox containing the local unix_full_system

Acceptance criteria

Below are the sub-tasks of this project:

• SIM RFC: Release Pipeline #3554
• SIM RFC: Initializing Binaries #3555
• SIM RFC: AI Selector Interface gl_runtime_selector.py #3556
• SIM RFC: AI Selector Backend #3557
• SIM RFC: AI Download Interface gl_runtime_installer.py #3558
• SIM RFC: AI Download Backend #3559

Blocked By