Right now exception are badly managed. They are typically caught close to their source, reported to the console and not propagated. We should distinguish between the exception that generate warnings, exception that somehow affect the workflow, and exception that abort TARDIS (if any), and properly handle them. This is not so simple given the multithreaded architecture of TARDIS, so some study on how exception propagation mixes with the executor framework is required before attempting such an endeavour.
Right now exception are badly managed. They are typically caught close to their source, reported to the console and not propagated. We should distinguish between the exception that generate warnings, exception that somehow affect the workflow, and exception that abort TARDIS (if any), and properly handle them. This is not so simple given the multithreaded architecture of TARDIS, so some study on how exception propagation mixes with the executor framework is required before attempting such an endeavour.