The new error handling framework should be used in place of all calls to fmt.Errorf() or Errors.new(). Using this framework will provide error codes to check against as well as the option to generate a callstack that will be appended to the error message when logging.error is set to debug in peer/core.yaml.
Using the framework is simple and will only require an easy tweak to your code. First, you'll need to import “github.com/hyperledger/fabric/core/errors” in any code that uses this framework.
Let's take the following as an example: fmt.Errorf(“Error trying to connect to local peer: %s”, err.Error())
For this error, we will simply call the constructor for Error and pass a component name, error message name, followed by the error message and any arguments to format into the message (note the component and message names are case insensitive and will be converted to uppercase):
err = errors.Error(“Peer”, “ConnectionError”, “Error trying to connect to local peer: %s”, err.Error())
For more critical errors for which a callstack would be beneficial, we can create the error as follows: err = errors.ErrorWithCallstack(“Peer”, “ConnectionError”, “Error trying to connect to local peer: %s”, err.Error())
If you are working with a team, try to agree upon a consistent component name to use. We may, in the future, add constants to allow searching for currently defined components for those using an editor with code completion capabilities; we are avoiding this for now to avoid merge conflict issues.
The display of the stack trace with an error is tied to the logging level for the “error” module, which is initialized using logging.error in core.yaml. It can also be set dynamically for code running on the peer via CLI using “peer logging setlevel error <log-level>”. The default level of “warning” will not display the stack trace; setting it to “debug” will display the stack trace for all errors that use this error handling framework.
Module core/errors/errors.go implements an error handling framework (package errors) supporting multiple languages.
From CallStackError interface (what can I do with a CallStackError?):
To instantiate an error implementing CallStackError interface (how do I get a CallStackError?):
Small examples are included in errors_test.go. It is also being used in peer/clilogging and peer/common.
A good introduction about errors and some potential ideas for error-handling in Go:
- https://blog.golang.org/errors-are-values
- https://blog.golang.org/defer-panic-and-recover
- https://github.com/mitchellh/panicwrap
- https://golang.org/src/runtime/traceback.go - how golang implements traceback
How to give examples?
- GOTRACEBACK=single is the default value for golang traceback information. This environment variable specifies level of detailed output generated when a Go program fails due to a panic or some runtime condition. The possible values for GOTRACEBACK are All,crash,system. It would be good to decide on what level of detail we might need to show for a traceback in Hyperledger by default.Along these lines do we need to have a configurable environment variable for the hyperledger specific stack trace which signifies the level of detail of the stack trace information?