Currently, if a call to a Ra server is unsupported, the Ra server will ignore the event and the call will eventually time out. This could happen when Ra servers sport different versions of Ra for instance.
It would be nicer if the Ra server would reply immediatly with an error to let the caller know about the actual problem.
How
The Ra server could use a reply effect in the catch-all clause, but that effect will crash the Ra server if the initial event is not a call (i.e. it doesn't have a From to reply to).
This patch introduces a new effect, maybe_reply, that has the same overall behavior as reply, but the lack of a From isn't fatal.
Why
Currently, if a call to a Ra server is unsupported, the Ra server will ignore the event and the call will eventually time out. This could happen when Ra servers sport different versions of Ra for instance.
It would be nicer if the Ra server would reply immediatly with an error to let the caller know about the actual problem.
How
The Ra server could use a
reply
effect in the catch-all clause, but that effect will crash the Ra server if the initial event is not acall
(i.e. it doesn't have aFrom
to reply to).This patch introduces a new effect,
maybe_reply
, that has the same overall behavior asreply
, but the lack of aFrom
isn't fatal.The returned error is: