returning different error in -[HTTPNetworkTransportRetryDelegate networkTransport:receivedError:for:response:retryHandler]

[RolandasRazma]

The problem I'm facing:
doing apolloClient.fetch()
request fails for reason Y (getting 401 to delegate)
trying to update token
token update fails for reason X
calling retryHandler(false)
request fails and callback in apolloClient.fetch() see error Y, where real error is X

What I would like to do is to modify (add new one and deprecate existing one) -[HTTPNetworkTransportRetryDelegate networkTransport:receivedError:for:response:retryHandler]
where you could return different error if you don't want to retry request

does that make sense @designatednerd or maybe there is better way to do that?

How it works now:

// HTTPNetworkTransportPreflightDelegate

func networkTransport(_ networkTransport: HTTPNetworkTransport, willSend request: inout URLRequest) {
  oauth.authenticate(&request)
}

// HTTPNetworkTransportRetryDelegate

func networkTransport(_ networkTransport: HTTPNetworkTransport, receivedError error: Swift.Error, for request: URLRequest, response: URLResponse?, retryHandler: @escaping (Bool) -> Void) {

  guard
    (response as? HTTPURLResponse)?.statusCode == 401 // Unauthorized - meaning token is probably expired
  else {
    retryHandler(false)
    return
  }
  
  oauth.updateToken { updateTokenError in
    retryHandler(updateTokenError == nil)
  }
}

// App

apolloClient.fetch(query: ...) { (result: Result<GraphQLResult<Query.Data>, Swift.Error>) in
  // Error is 401, and it is correct from apollo perspective, but its not real error if oauth.updateToken fails for whatever reason - lets say internet connectivity problem
}

What I would like to do is:

// HTTPNetworkTransportRetryDelegate

func networkTransport(_ networkTransport: HTTPNetworkTransport, receivedError error: Swift.Error, for request: URLRequest, response: URLResponse?, continueHandler: @escaping (ContinueAction) -> Void) {

  guard
    (response as? HTTPURLResponse)?.statusCode == 401 // Unauthorized - meaning token is probably expired
  else {
    continueHandler(.fail(error))
    return
  }
  
  oauth.updateToken { updateTokenError in
    if let updateTokenError = updateTokenError {
      continueHandler(.fail(updateTokenError))	
    } else {
      continueHandler(.retry)
    }
  }
  
}

// App

apolloClient.fetch(query: ...) { (result: Result<GraphQLResult<Query.Data>, Swift.Error>) in
  // Error would be updateTokenError, in internet connection problem it would be "fail to reach host" instead of 401 what came from apollo
}

I think this could be done without breaking changes by forwarding to existing method if new one is not implemented

[designatednerd]

So it sounds like the error from the retry is not actually being propagated to the original call is the biggest issue here - is that correct?

[RolandasRazma]

no, because my token update fails I know that there is no point in retrying, in delegate say "don't retry" - everything works correctly.
Except that logically I have different error (from token update) and not from apollo, but I have no way of propagating it up the chain

[RolandasRazma]

updated ticket with code samples, maybe will be easier to understand

[RolandasRazma]

would you accept such PR?

[designatednerd]

Thanks for adding the code samples, that does help. Yeah, I think this would be reasonable. Would you like to take a stab at a PR on this?

[RolandasRazma]

yes, will do that

[designatednerd]

This has shipped with 0.26.0!