Bacalhau project report 20220502

This week we have focused on the process of incrementaly refactoring the existing codebase whilst implementing the "run a docker image" feature.

The main feature of the work has been to define what the "compute" interface actually looks like and most importantly, how the different types of storage are handled by different implementations of the "compute" interface.

Things we have implemented in the "docker-image" branch:

• generic scheduler library (so writing new schedulers is much easier now)
  • unit tests for scheduler using "inprocess" scheduler
• compute engine interface implementation
  • noop comoute engine implementation (so we can have unit tests with mocks)
• configuration module (so we can configure the daemon via cli flags, config file and/or env variables)
• compute engine now "maps" storage engine implementations
  • so for example "ipfs" means something different to the WASM compute engine than it does the docker engine
  • this means we can implement "mount" based storage engines seperately from "lambda" based ones
• everything has been moved into the "pkg" folder under an appropriate package