An attempt to add causal dependency to the replay scheduler. To use this, set_target_clock should be invoked on a ReplayScheduler instance with the clock of the failure event (provided as a slice of integers). Some things to resolve still:
[ ] To make this work, schedulers now receive references to Tasks rather than just TaskIds. I think we discussed that schedulers should generally have access to more information about tasks, so if this is the approach we take, the Task interface might deserve some more cleanup. Even more so if schedulers are eventually intended to be defined in a separate crate.
[x] (minor) Any RNG steps in a schedule following a skipped task should also be scheduled. Need to add a test for this and implement.
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
An attempt to add causal dependency to the replay scheduler. To use this,
set_target_clock
should be invoked on aReplayScheduler
instance with the clock of the failure event (provided as a slice of integers). Some things to resolve still:Task
s rather than justTaskId
s. I think we discussed that schedulers should generally have access to more information about tasks, so if this is the approach we take, theTask
interface might deserve some more cleanup. Even more so if schedulers are eventually intended to be defined in a separate crate.By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.