Rodney, Yes, it's only meaningful when the Map contains one-to-one mappings.
But then, having a BijectionMap doesn't quite address my specific use cases. In my case, I got this HashMap from a routine (not written by me). I know it's bijection and I need to perform inverse hash lookup. That's why I need a key-value swapping routine. If I use a BijectionMap, I have to copy the value from the original map into the bijectionMap before doing the swapping. It doubles the amount of work. It's an overkill... Having said that, however, I do see some interesting applications beyond my (selfish :-) use cases. For instance, it seems to me this BijectionMap can be (semantically) the supertype of the existing DoubleOrderedMap. At 08:17 pm 19-09-2002, you wrote: >It seems to me this is only meaningful when (or at least, this is only a >correct implementation when) the Map is a bijection--with a one-to-one >mapping between keys and values. Otherwise some keys in the source map will >not appear as values in the destination map, and which keys disappear will >be dependent upon the order they are returned by the iterator. > >I don't know what your use case is, but I think I'd rather see an explict >bijection class (extends Map) that supports this functionality than a >MapUtils method that is only meaningful for a subset of Maps. > >What do you think? -- John Yu Scioworks Technologies e: [EMAIL PROTECTED] w: +(65) 873 5989 w: http://www.scioworks.com m: +(65) 9782 9610 -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
