transport_failure_callback gives no information on failure reason

[Zoetermeer]

All we are given in this callback is the hash version of the event we originally intended to send, but it would be useful to also be given the exception that caused the send failure as an additional argument. Otherwise we have no idea why the request failed in the first place.

[EtienneDepaulis]

Hello @nateberkopec I would be happy to help on this issue as it impacts us.
Is a PR on the current the best approach for you ?

[nateberkopec]

@EtienneDepaulis Yes please!

[EtienneDepaulis]

I've finally took the time to look at the code and I as I'm no ruby expert I would like to have your opinion on the best approach:

Here is the code responsible for calling the transport_failure_callback proc:

def failed_send(e, event)
  @state.failure
  if e # exception was raised
    configuration.logger.error "Unable to record event with remote Sentry server (#{e.class} - #{e.message}):\n#{e.backtrace[0..10].join("\n")}"
  else
    configuration.logger.error "Not sending event due to previous failure(s)."
  end
  configuration.logger.error("Failed to submit event: #{get_log_message(event)}")
  configuration.transport_failure_callback.call(event) if configuration.transport_failure_callback
end

We are interested in the properties of the error raised passed to this method by e. The problem is that transport_failure_callback expects only one param and doing something like configuration.transport_failure_callback.call(event, e) would be a breaking change.
But as mentioned, this feature is supposed to be introduced in v3, so those kind of changes are acceptable.

Another option would be to store the exception in the context of the event (extra section).

Which approche seems better for you @nateberkopec ?

[nateberkopec]

@EtienneDepaulis Open a PR with a breaking change against the v3 branch, please.

[st0012]

close with #1003