I've just made a simple bean with

setStringArray(String[] arr);
setProperties(Properties props);

in combination with

<bean...>
  <property name="stringArray">
    <value>one,two,three,four</value>
  </property>
  <property name="properties">
    <value>
    # inline properties file:
    test.one = yes
    test.two = no
    </value>
  </property>
</bean>


then I wrapped the configured bean in a BeanWrapperImpl and found the
following to be counter-intuitive:


1/
wrapper.getPropertyValue("stringArray[2]") returns "three" (no surprise)
BUT: wrapper.isReadable("stringArrray[2]") returns false?


2/
wrapper.getPropertyValue("properties[test.two]") returns "no" (no surprise)
BUT: wrapper.isReadable("properties[test.two]") returns false?


briefly browsing the cvs the reason for the mismatch between the two methods seems to come from how they lookup their propertyDescriptor.


line 993:
PropertyDescriptor pd = getPropertyDescriptorInternal(propertyName);



versus lines 531-538:
public Object getPropertyValue(String propertyName) throws ...{
    BeanWrapperImpl nestedBw = ...
    String[] tokens = getPropertyNameTokens(getFinalPath(nestedBw, propertyName));
    return nestedBw.getPropertyValue(tokens[0], tokens[1], tokens[2]);
}

protected Object getPropertyValue(String propertyName, String actualName, String key) throws BeansException {

    PropertyDescriptor pd = getPropertyDescriptorInternal(actualName);
    ...

notice how actualName is used here and not propertyName!



it's probably fair to suspect an equal mismatch between isWritableProperty(...) and setPropertyValue(...)?