Record execution order
joealcorn opened this issue · comments
When #13 was shipped we started running control & candidates in a random order. This is in order to weed out any issues caused by hidden dependencies between them.
However we never record the execution order anywhere, which could be a helpful piece of information when debugging these issues, so we should probably start doing that somehow.
How would you like to record them? Would logging be sufficient?
For example:
logger.debug('Executing control: {}'.format(control_fn.__name__)
logger.debug('Executing candidate: {}'.format(candidate_fn.__name__)
The library could also pull the fully qualified name as well, if that'd be more helpful. What do you think?
I don't think logging is enough - it should be available on the Result
instance so that it's available during publishing.