Why do scope modifiers on fields in domain classes prevent validation?

Question

I'm just starting with Grails (coming from Rails) and I noticed that Grails really doesn't seem to like scope modifiers on fields in domain classes.

I had understood that all unscoped fields in a domain class were by default public, but if you actually declare it public, Grails won't validate it.

class Person {
  public String firstName
  public String middleName
  public String lastName
}

If you add a constraint, Grails will throw a NotReadablePropertyException exception when you call validate()

class Person {
  public String firstName
  public String middleName
  public String lastName

  static constraints = {
     middleName nullable: true
  }
}

However if you take out the public declaration, everything works normally.

Can someone explain what's going on behind the scenes with the scoping in domain classes? Hard to understand why explicitly declaring something public which is already public would break the framework. I'm guessing you wouldn't want to declare anything 'private' either, although it would be nice if there was away that a fields which shouldn't be manipulated directly could be hidden from consumers of the domain class.

Answer 1

When you add a field to a Groovy class without a scope modifier, it's more that it's inferred to be public than being actually public. The compiler converts the field to a private field and generates a public getter and a setter for it, although it won't overwrite a getter or setter that you wrote. This is convenient because you can later write getters and/or setters to implement business logic and not affect the callers.

But a public field (declared as 'public') is just that - a public field. There's no generated getter or setter. I recommend using a decompiler to see this in action - create a simple POGO in src/groovy, eg

class Thing {
   String realProperty
   public String fieldButNotProperty
}

and open up the .class file with http://jd.benow.ca/ or another decompiler.

GORM automatically assumes that typed properties are persistent unless you exclude some with the transients list. The type is required so it knows how to persist the data, and properties like def name will be ignored. Properties in this sense are similar to JavaBean properties - a matched getter/setter pair.

Hibernate has no support for Groovy and doesn't know what's going on under the hood - it just calls your getters and setters to set and access field data during persistence. So the Groovy compiler adding those in makes it easy for POGOs in Grails to be persisted by Hibernate. And you could do this yourself - add in a getter and setter with correct names and data type (eg String getName() and void setName(String name) and it will be treated as a persistent property, even if you do nothing with the values.

The reason for the NotReadablePropertyException is that there's no getter to call for your 'property'. Even though your fields are perfectly accessible, you've effectively hidden them from GORM and Hibernate.

Answer 2

If you add a constraint, Grails will throw a NotReadablePropertyException exception when you call validate()

Never noticed this before, sounds like a bug

it would be nice if there was away that a fields which shouldn't be manipulated directly could be hidden from consumers of the domain class.

If you want to prevent direct access to a property, simply add a getter and setter. In the (contrived) example below, I ensure that name is always read/written as an upper case string.

class Person {
  public String firstName
  public String middleName
  public String lastName

  public void setFirstName(String name) {
    this.firstName = name.toUpperCase()
  }

  public String getFirstName() {
    return this.firstName.toUpperCase()
  }
}