When using the go-sql-driver it feels like I'm re-inventing the wheel a little bit. I'm used to languages where you have a data model (like a User class) which is strictly related to the database model. Then you can have additional models for what you want to expose on an API of course, but still, you have a model for the database object. In Go I'm not sure how to do this the best way. Below is a GetUserByEmail function.
func GetUserByEmail(email string) (*myapp.User, error) {
smt, err := database.Prepare("SELECT * FROM users WHERE email = ?")
if err != nil {
return nil, err
}
rows, err := smt.Query(email)
defer rows.Close()
if err != nil {
return nil, err
}
var users []*myapp.User
for rows.Next() {
var id string
var confirmed bool
var password string
var email string
var updated string
var created string
err = rows.Scan(&id, &confirmed, &password, &email, &updated, &created)
if err != nil {
return nil, err
}
user := myapp.User{id, confirmed, created, password, email, updated}
users = append(users, &user)
}
if users == nil || len(users) == 0 {
return nil, errors.New("User not found.")
}
if len(users) != 1 {
return nil, errors.New("Nr of users fetched: " + string(len(users)) + ". Expected: 1.")
}
return users[0], nil
}
One problem here is that if I change the order of password and email in the database, my application will silently switch those and I will get no error. That's an awful and dangerous behavior. Can Scan() take keys somehow or can I do this in some other way? I could of course look at Columns() and take the index order from there as an int (although Go doesn't seem to have that function built in) and map it to my User variables, but do I really need to do all that for all my database methods? Is it abstracted somewhere or do I need to do that myself?
