moose
moose copied to clipboard
Remove recover testing from all Steady tests
Reason
The way these tests go is:
- regular solve
- recovers after it, exits immediately after setup with
executioners/executioner.test_steady_adapt:
executioners/executioner.test_steady_adapt:
executioners/executioner.test_steady_adapt: Cannot recover steady solves!
executioners/executioner.test_steady_adapt: Exiting...
executioners/executioner.test_steady_adapt:
there's some testing going on in that it shows we can initialize a new simulation. But it s really not a complete picture like running a transient recover.
Design
Find every test running a steady solve, add recover = false to the test spec OR Do that but with a python script OR Modify the test harness to detect Steady and skip by itself (best solution)
Impact
Less testing, less resource use, lower turnaround on civet
Should recovery be allowed for Steady in the code?
I think so if we are trying to recover in between nonlinear solves for example.
I don't know that we can do that