Good catch. . .once I come up for air, I'll definately committ it.
Also, have you added the test that caught this in the patch?
David
Zhong ZHENG wrote:
Hi, I found another bug in Pluto 1.1 branch that breaks the portlet spec.
The portlet spec (xci, SPEC 91, PLT 14.1) states that:
The getMap method returns an immutable Map of String keys and String[]
values containing all current preference values. Preferences values must
not be modified if the values in the Map are altered.
I added a checkPreferencesValuesNotModified() method in the testsuite
class SimplePreferenceTest, to check PLT 14.1. That test failed: In
fact, if I changed the values of the string array in the map returned
(NOT to put an entry into the map, since that operation is not allowed),
the preferences values are changed accordingly.
The bug resides in the class PortletPreferencesImpl. I updated the
method getMap() as the following, and then the spec was respected:
public Map getMap() {
Map map = new java.util.HashMap();
Iterator it = preferences.keySet().iterator();
while (it.hasNext()) {
PortletPreference pref =
(PortletPreference)preferences.get(it.next());
map.put(pref.getName(), pref.getValues().clone()); // <==
Here we clone the string array.
}
return Collections.unmodifiableMap(map);
}
I submitted a patch (SVN diff file) for PortletPreferencesImpl, as well
as a SVN diff file for the updated test class SimplePreferenceTest, on JIRA:
http://issues.apache.org/jira/browse/PLUTO-167
Hope that will be applied to SVN.
Regards.
--
ZHENG Zhong
1 Avenue Alphand
75116 Paris, France
+33 6 76 80 45 90
heavyzheng {AT} gmail {D0T} com
http://heavyz.sourceforge.net
http://heavyz.blogspot.com
http://spaces.msn.com/members/zhengzhong
