How do solver processes work
Table of Contents
updateAndCompensate¶
def updateAndCompensate(applied_rule, movement_list):
is the key method to consider in order to get stable expands in ERP5 simulation. The key is to provide an appropriate movement_list.
Example 1:
- simulation movement is forced (Adopt Forced Decision) to new start_date: t2
- no backtracking happens (ex. no Postpone)
- snapshot keeps of copy of t1, original start_date
- next time expand happens, t1 value is used in the movement_list passed to updateAndCompensate
- as a result, the expand is stable
This example shows that the concept of taking snapshots helps reaching a stable expand (when no backtracking happens).
Example 2:
- we take the same example
- but the parent order is postponed
- movement_list will either lead to compensate (ie. postpone) simulation or to update simulation
This example shows that the concept of snapshot is not useful to prevent divergences when forward expand happens. We have nothing to say "do not touch this property, it is like this, it is not
divergent".
Related Articles¶