Stevan Little wrote:
>> Hmm, perhaps it would be a start for a WTF entry?
>
> Sure, patches welcome, or stop by #moose for a commit bit.
I'm attaching a patch. This is my first patch, done with 'diff -u', which I
hope is good enough. But if there is something wrong, please bear with me
and tell me what else would be appropriate.
Bernardo
--- /home/brb/localperl/lib/site_perl/5.10.0/Moose/Cookbook/WTF.pod 2007-12-31 11:46:06.000000000 -0500
+++ WTF.pod 2008-05-19 13:28:20.000000000 -0400
@@ -109,6 +109,26 @@
return reverse @rv;
};
+=head3 Why do the accessor modifiers of a superclass' attribute stop working after redefining or extending the attribute in a subclass?
+
+When an attribute is redefined or extended in a subclass, a completely new
+attribute accessor is created that overrides the original one. This happens
+through two mechanisms: if an attribute is redefined (with C<has 'some_attribute'>),
+everything related to the attribute (accessor methods, etc.)
+is generated anew for the subclass; if an attribute is extended (with
+C<has '+some_attribute'>), the corresponding superclass' attribute is first cloned,
+and then the requested extensions or modifications are applied to the clone.
+
+Method modifiers are applied specifically to methods in the class in which they
+are originally defined, and are not re-applied for each subclass. It doesn't
+matter whether a method was written by the user explicitly or by an attribute
+automatically, and so it will do nothing special if it's applied to an accessor.
+
+When redefining or extending an attribute, the code refs of the new accessors are
+different from those of the superclass. The modifiers of the superclass apply to the
+superclass' code refs, and unless new modifiers are defined for the subclass, no
+modifiers are applied to the subclass' code refs.
+
=head2 Moose and Attributes
=head3 Why don't attributes I inherited from a superclass work?
