Coding Practice for Bound Property Names

Question

I think this simply an issue over coding practices, but if I am right then all the tutorials and documentation on how to write Beans are promoting poor coding practices. Speicifcally, it bugs me that String literals are always used in all the code I see. As in the following code:
class MyBean  {
   public void setValue(value newValue) {
      beansupport.firePropertyChange("value", oldValue,  value);
}the practice that I have starting using is:
class MyBean  {
    public static final String propertyValue = "value";
   public void setValue(value newValue) {
      beansupport.firePropertyChange(propertyValue, oldValue,  value);
}This way, a PropertyChangeListener can refererence MyBean.propertyValue instead of using the literal "value" -- eliminating the chance for error and also there don't have to be multiple literals floating around using up space. Is there any reason this isn't commonly accepted practice or did I miss something?

Answer 1

Strings appearing in source code can be reduced to a constant, like an enum. So it is likely that the code with strings is actually faster, as the value is known before compilation. It is easier to read. Keep in mind that strings get interned every now and then and thus get reduced to this equivalency:
public method blah( "value", some stuff )...
public method nah(*){ do stuff; let x = "value"; more stuff with "zoiks!" }
ends up like
key k1 = "value"
key k2 = "zoiks!"
what the jvm sees;
public method blah( key1, some stuff )...
public method nah(*){ do stuff; let x = key1; more stuff with key2 }
The string values get "inlined" and can be checked during compilation, allowing for many optimisations not otherwise possible.
Hope that helps a little.
Andrew