I have struct that have a method that serve the response.
type ctrl struct {
*base.AjaxCtrl
file ini.File
}
func (rcv *ctrl) getVar() string {
return mux.Vars(rcv.Req)["location"]
}
func (rcv *ctrl) getFile() string {
return location.JoinPaths(folder, rcv.getVar()+ext)
}
func (rcv *ctrl) upload() {
file, err := ini.LoadFile(rcv.getFile())
if err != nil {
rcv.AddErr("TextError", err.Error())
return
}
rcv.file = file
}
// Convert text to json structure
func (rcv *ctrl) convertToJson() string {
js, err := json.Marshal(rcv.file.Section("text/signup"))
if err != nil {
rcv.AddErr("ConvertError", err.Error())
return ""
}
return string(js)
}
func (rcv *ctrl) serveHttp() (types.SuccsJSON, types.ErrorsJSON) {
rcv.upload()
if rcv.AnyErrors() {
return nil, rcv.Errs
}
str := rcv.convertToJson()
if rcv.AnyErrors() {
return nil, rcv.Errs
}
return c.Sucss, nil
}
The method serveHttp() handle the response to client. As you can see, I am handling the error with the method AnyError() everytime. I find this way is pretty boring and maybe wrong design. Would it be better to throw a panic instead of error handling?
func (rcv *ctrl) upload() {
file, err := ini.LoadFile(rcv.getFile())
if err != nil {
rcv.AddErr("TextError", err.Error())
panic()
}
rcv.file = file
}
No, it would not be better. Checking your errors is not bad design in Go. Panicking when there is no reason to do so, on the other hand, is bad design, because it's harder to test and debug.
The Go Wiki explicitly discourages the use of panic outside of truly exceptional cases, when the error is unrecoverable and state of the program can't be fixed.