search

mle-infrastructure / mle-toolbox

Lightweight Tool to Manage Distributed ML Experiments 🛠
https://mle-infrastructure.github.io/mle_toolbox/toolbox/
MIT License
3 stars 1 forks source link

Better error handling for CLI #19

Closed RobertTLange closed 3 years ago

RobertTLange commented 3 years ago

We need better error handling for invalid command line input. When launching an experiment without the --no_protocol option there is a set of fixed steps:

  1. Greet
  2. List experiments from protocol overview
  3. Ask if user wants to delete [e_id/N]
    • If yes: Ask user if they want to delete another experiment
    • If no: Ask for purpose of this new experiment

  4. Run experiment

    Currently an experiment is aborted if in step 3 the user doesn't provide the right form of input. E.g. if an e_id doesn't exist in the database or if one accidentally types the purpose when once is asked for an e_id. We should make this more robust and verbose.

RobertTLange commented 3 years ago

Addressed in 621b90b522215487caa3ad61ee09e552ac937dd0 - now the loop of asking for [e_id/N] is continued after a message ala "The e_id is not in the protocol db. Please try again: [e_id/N]" is printed out.