Implement resume or re-start of build from defined/failed repository

[mareknovotny]

It would be very helpful to have RESUME/RE-START feature in case of CLI usage for debugging and fixing the problem.
Right now you need to run only whole Build chain repository list or manually write command for skipping succesful repository builds.

Tasks

Beta Give feedback

1. Implement resume such that for each cli build it stores a state file. The resume command will pick up and continue the build from this state file #450
   +0
   
2. Add verification whether the stored state actually reflects the current state of the workspace #452
   +0
   
3. Add option to specify start project instead of just resuming from the first point of failure #455
   +0
   
4. Add option to re-checkout a project or list of projects #457
   +0
   
5. Override definition file or part of the definition file (TBD. Refer Add verification whether the stored state actually reflects the current state of the workspace #452)
6. Add option to create multiple state files so that user can select from which state they want to resume

[Ginxo]

• metadata
• all the parameters from BC
• df: URL
• rest of the parameters
• definition-file: the computed definition-file
• checkout
  • project1: true
  • project2: true
  • project3: false
  • project4: true
• execution
  • projectX
    command: X

let's use BC execution folder as the workspace to store .state.json file

-state flag either on build and resume functionality -state kike -> .state.kike.json

[shubhbapna]

One interesting thing to explore would be serialization of the config service. If we are able to serialize the config service and store in the .state.json file and deserialize it when we resume, then we won't have to manually figure which meta data to store