This is a prototype I've made for @svandenhaute but maybe also of interest to Micha Pfeiffer

This prototype allows a retry handler to decide that on a particular task failure, the task should complete with a value of its own choosing.

In the scenario I talked of with @svandenhaute this would allow a tasks exception to be replaced by a substitute value that later other tasks would be able to observe - rather than task flow stopping - and so in a sense allows reification of the exception in a regular value.

In Micha's case, this may be a route to "permanently failing" a task into the checkpoint database, so that it does not re-run when restarting from checkpoints. (or might not - because this code means that other parts of a workflow will no longer observe a failure)

The supplied test case shows catching a particular exception and making that task not fail but instead complete with the value 8.

I'm unclear how this works in practice but the prototype was quick to make and I appreciate feedback.

Changed Behaviour

none

Type of change

• New feature