Redo error message with new log dispatchers

[joasode]

No description provided.

[Chroxvi]

I like this idea. I think it provides a significant improvement over the existing error handling. However, I would like to propose a slightly different implementation strategy:

• Use the cotainr.container.logger for logging the FATAL message instead of the subprocess logger.
  • Get the verbosity level from SingularitySandbox._verbosity.
• Instead of directly calling sys.exit(0) directly in the container module, raise a RuntimeError from the CalledProcessError and catch it in the relevant place in the cli module, then call sys.exit(0) there as the handling of the RuntimeError. (Calling sys.exit(0) outside of the cli module is a bit brutal since people may (theoretically) what to just use cotainr as a library and build their own user interface for it.) (Writing this, I realize that we may have a problem here:

  cotainr/cotainr/pack.py

  Line 276 in 4855944

  sys.exit(0)

  )

Also we might consider:

• Adding extra cotainr DEBUG messages around this call to the subprocess to:
  • make it easier to identify where we are in the cotainr code base when the problem occurs
  • record the value of some of the key variables we are passing around in cotainr related to the subprocess
• Including the cmd in the FATAL log text even when verbosity<4.