Add an ugly fix to avoid endless disco#info queries with each message (with receipts)
We need to check if the remote entity supports 0184, but if it doesn’t support disco#info, then we will get an iq type="error" and nothing will be cached, leading to disco#info queries being sent each time. Keep a cache valid 2 hours of the JIDs which replied with an error. TODO: check that this the kind of time period we want.
Showing with 103 additions and 3 deletions