Closed brong closed 2 years ago
I can see that the important detail here is that the mismatched replica mailbox has never had a message, and that "it had messages once, but they've since been deleted" should still cause the replication to abort, rather than overwriting the mailbox. It might be nice to have a test for that specifically, to protect against future regressions and as documentation of our expectation.
I'll put another test for that case. Good point!
Tests for https://github.com/cyrusimap/cyrus-imapd/pull/3608