The Slipstream docs have a section on retries for establishing a connection.
But I don’t see anything about retries for messages. The docs for push/5 show using handle_reply/3
to handle an ack, but suppose an ack never comes, or takes too long to come. I think that could happen if:
- The server had a bug and lost the message
- The server channel crashed
- The TCP retries timed out and the connection crashed
I think in cases 2 and 3 some kind of error would be raised. But in any case, am I correct in assuming that the answer for message-level retries is “roll your own, according to your needs”?