from: Phil Steitz <[EMAIL PROTECTED]>
1. I made no attempt to handle or mask nulls -- puts with null keys will throw NPEs.
This seems harsh. I would prefer to support nulls if possible.
OK. No problem.
Could be I was being too lazy. Since toLower() is "idempotent" (calling it repeatedly has same effect as calling it just once) it "works" to just override convert(), since in AbstractHashedMap, get, put, remove, containsKey, all start by converting the key (nice!). All I did was to extend HashedMap and override convert to return key.toString().toLowerCase(). The effect is that you actually store all lower case strings as keys (or null, once I fix this). The behavior should be the same as the impl that David Graham posted with PR #24537 (unless I am missing something -- quite possible).
2. I made no attempt at allowing the locale to be set or otherwise changing the default behavior of toLower() in convert().
I'm wondering how your implementation works. I was expecting the isEqualKey() method to be overridden, rather than converting the key on add. (Note that the same code also has to go in a MapEntry)
3. I need the secret recipe for making the "canonical serialized object file" so that I don't have to dummy out the "compatability" tests (or not :) Is this documented somewhere?
Its commented out in the bottom of some of the tests. (HashedMap?)
Ah...I was wondering what that cruft was about ;) Thanks.
Phil
Stephen
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
