openQA Project (public)

First, in theory it should be expected that the openQA API route returns a list of ids. And the jq command is already written in a way that it would walk over multiple ids. The assignment to the bash variable would have list instead of a single job then. But the "parse error" sounds like openqa-client --json-output would not actually output json output.

Multiple things we could make out of this:

• Find out why openqa-client --json-output should output a plain id list (let's start with that)
• Fix the actual issue in https://github.com/os-autoinst/os-autoinst-distri-openQA/
• Suggest to use serial terminal by default for terminal tests
• Make sure serial terminal based tests still have a nice video (or at least exclude from the video)

Geez, now I see the relation. It's a recent regression. Please fix urgently

revert was merged. That is making the issue less severe. The ideas we discussed were that we could improve the logging. Underlying libraries should not log directly but be introspectable, e.g. provide the current configuration to be read by applications. Applications should be quiet by default, only output error messages. Log messages should be written to stderr rather than stdout. Higher levels of verbosity could be enabled with a -v|--verbose command line parameter or allowing multiple levels like is done in https://github.com/os-autoinst/openQA/blob/master/script/openqa-label-all#L27