PietroPasotti

A precedent for the pattern of calling an exit function instead of raising an exception would be pytest's `xfail()`, which I like a lot. But I don't mind asking the...

> Given those points, I tend to think we should move in the opposite direction of making it normal for the charm to expect multiple events. (We _could_ still decide...

now that [`snapd-control` has been accepted for jhack](https://forum.snapcraft.io/t/snapd-control-request-for-jhack/35432/2), this could go away after we: - add a snapd-control slot to jhack - catch this error and recommend connecting that slot,...

Workaround for SOME issues (`juju ssh` compatibility) is to use jhack from sources (not the snapped version). Commands that rely on juju cli output may still break, as some of...

todo: - derive PVC name from kubectl so we can drop the find hackery - mount the separate container mounts into one another so that dirs don't appear empty (using...

I'm working on this, I think the issue has appeared on some recent juju version (did the permissions change on the charm files at some point I recall?) the importerror...

picking this up as it breaks TLS integration

I've taken a look at the context in https://github.com/canonical/operator/issues/970 and IIUC we don't need to check for CMR because even for model-internal grafana-source relations, we still need ingress URL as...

if not we can use something like ```python # cfr: https://github.com/canonical/operator/pull/1096/files # cfr: https://github.com/canonical/operator/issues/970 def _is_cmr(relation: Relation): """Temporary workaround for checking if the given relation is a cross-model relation. Note...

if we do this, TLS stops working.