A DESCRIPTION OF THE REQUEST :
Since LHM is ordered you can use LHM.iterator() LHM.values() and LHM.entries() to get items in order.  However, you cannot get values in *reverse* order as there is no equivalent of Vector.listIterator(int).

JUSTIFICATION :
Performance and  usability.

This is what i expected it to do already !

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
LHM.values() and LHM.entries() to return an AbstractSequentialList rather than a Collection interface (implemented as an anonymous extension of AbstractCollection).  **This would not break any existing applications**
ACTUAL -
LHM.values() and LHM.entries() return an AbstractCollection that cannot therefore present a ListIterator to the user.

CUSTOMER SUBMITTED WORKAROUND :
Currently you have to use :

Collections.list( Collections.enumeration( linkedhashmap.values() ) );

The performance of this is appalling considering the underlying collection is a linked list.  Obviously, we maintain a second index to get around this.
  xxxxx@xxxxx   2005-05-06 06:41:27 GMT

N/A

We should seriously consider the same kinds of bidirectional capabilities for
LinkedHashMap as are currently being added to LinkedList, etc.
Posted Date : 2005-08-08 00:25:26.0

Part one of the proposal is to modify LinkedHashMap.values() (both signature and implementation) to return a List.  I can't comment on the utility of this change, but I can think of nothing fundamentally wrong with the idea.

Part two is to modify the "entries()" method; though this method doesn't exist, only entrySet(), which clearly can't return a Set.  But we can interpret this as a proposal to create a new method, which I would call entryList(), returning the same data as entrySet() only as a List.

Again, I don't see why not.  All the issues I've thought of have only to do with add() operations which these collection views don't support anyway.

Note that either of these changes would have the effect of promoting LinkedHashMap to become a bona fide _type_ in the API, no longer a mere implementation class.  Again... I don't see anything wrong with that.

This bug is related: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4848853.