VL scaling support for Simulation #1961
Conversation
|
Violations seem not to be removed anymore when setting "remove-violations-after:" in the punishments.yml config. Could be something i messed up. |
| private int getViolations(PunishGroup group, Check check) { | ||
| int vl = 0; | ||
| for (Check value : group.violations.values()) { | ||
| if (value == check) vl++; | ||
| if (value == check) { | ||
| vl = Math.max(vl, (int)Math.ceil(value.getViolations())); | ||
| } | ||
| } | ||
| return vl; | ||
| } |
There was a problem hiding this comment.
Violation entries added to PunishGroup.violations via handleViolation() are still getting deleted correctly, but the VL count returned by getViolations() no longer correspond directly to the amount of key-value pairs with the same value (check) as the supplied parameter, but is instead taken from Check.getViolations(), which would only decrease upon a Check.reward() call.
To fix this issue, there needs to be some sort of way to keep track of the VL gain from individual violations.
|
I would consider this PR not important enough to be merged upstream. |
This PR introduces a setting that allowed adding multiple VLs per Simulation flag based on the amount of advantage.
Some changes were made in
PunishmentManagerto support this.