Currently it's quite difficult to debug issues with inferno-ml-server. If something goes wrong during script evaluation, Inferno initialization, etc..., the only way to distinguish errors is by whatever output is produced by the implementation of displayException by RemoteError. Instead, we could serialize the errors and provide this in the errBody. Then any calling client could try to decode the body and get back the actual exception
Currently it's quite difficult to debug issues with
inferno-ml-server. If something goes wrong during script evaluation, Inferno initialization, etc..., the only way to distinguish errors is by whatever output is produced by the implementation ofdisplayExceptionbyRemoteError. Instead, we could serialize the errors and provide this in theerrBody. Then any calling client could try to decode the body and get back the actual exception