Open jku opened 1 year ago
So signing_event.update_targets()
could
This would force the delegated metadata signers to sign in the same signing event.
However, I'm not sure this is appropriate in every case: what if the delegated metadata is already invalid (e.g. expired) and the signers are not available? This would prevent making changes to the delegation in that situation... This might be acceptable as long as this really applies only to delegations that have changes -- this way changes to delegating metadata can't be blocked because delegated signers are not available: so we should not just test the validity of all delegated roles when delegating metadata is changed, we should only test the ones whose delegation was changed
Some more notes after testing a bit:
.signing-event-state
, and the repository and other signers consider the signing event incomplete until the invites are accepted). This covers most cases of changing delegations but not all.signing-event-state
file is no longer needed at all
I think currently if a role (delegation) exists, and gets a change that makes the current delegated metadata invalid but does nothing that actually changes the metadata (like threshold is raised with no other changes) , then our tooling does not notice that the repository has invalid metadata in it.
What should happen is that signing event notices the metadata is invalid, and creates a new metadata version for that delegated role in the same signing event: this would force re-signing the delegated metadata as well