Open shigmas opened 9 years ago
For error handling, propose adding API to set an error handler on the network layer, so we can show whatever error we receive to the user.
Also, on startup, we should just quietly give up for the time being.
This is going to be an uber bug. This requires work on the server to provide some more information. On the client, we need to start handling those errors. And context determines importance in some cases. So, this will be ongoing, but we can close this when we get the general case, and some specific ones, and open issues for specific ones.
When it can't connect to the logging service, it starts spewing out requests (that get sent when the connection is restored).
When there's no connection, we should just notice that, and give up. When service is restored, we can sync back up. (Meaning, we should have a good "see what happened since we were last connected" kind of call.