On Wed, Jan 09, 2013 at 06:39:40PM -0800, Dimitre Novatchev scripsit:
> Graydon,
> 
> I am glad this was useful.
> 
> Forgive me for not mentioning keys explicitly -- I can't imagine that
> someone wouldn't use keys for this :)

Once one has figured out that the problem is one of relative frequency,
not something to do with determining uniqueness, yeah, keys are pretty
obvious.

_Before_ that happens, well.  Rather less obvious!

-- Graydon