The following topic Proposition de supprimer la notion d'identité désactivée mais non révoquée et la notion d'adhésion is asking a question of optimization:
Is it better to schedule a dispatch and cancel it later or just rely on external mechanism for automation?
Example: schedule membership expiry in case of inactivity / schedule certification expiry / schedule revocation… and cancel if user was active / renewed the cert / met a certain requirement in the meanwhile…
The advantage of the scheduling approach is that everything relies on blockchain and that the rules we choose do not depend on someone applying it. The advantage of the external worker is that it is lighter both in terms of storage and blockchain computing power.
My general feeling right now is that we should choose the scheduling since it is more straightforward and requires less external design. If we reach the maximum of our blockchain capacity we might want to reconsider this and migrate to external worker, but this should only arrive at a point where Ğ1 is large enough that it can fund these kind of developments easily.
This subject is not as “easy” as the one about the production of UD (Production du DU) because the gain in performance is not as obvious and because the membership and certification expiry have more consequences than the claiming of UD.