A new XML regression test has been checked in, which runs a plethora
of combinations of getters, setters, and adders, across a range of
different objects, and attempts to 'throw off' the mapping system.
A number of bugs have fallen out of this; two of them, IMO, are
critical blockers before ship.
CASTOR-1121: if a set-method="method" exists, at minimum on a
boolean, and that boolean is false, the set method is never called.
CASTOR-1122: Deleted. Ignore me.
CASTOR-1123: Not a killer, but strange: add() methods on a hashmap
or set do not attempt to set the key/value pair - just the value.
CASTOR-1124: Add methods are 100% non-functional. End of story.
It's all covered by the regression test, commented where I ran into
the problems in the various set/validate methods, for your perusal.
IMO, we're at a stop ship until castor-xml gets this sorted.
This test will not run, even through decorative commenting as I have
tried, until 0.9.6 head is fixed, or the multiple-setter changes are
backed out.
