reg/unreg testing

[Tom Anderson]

From: "David Relson" <relson at osagesoftware.com>
> Looking at these results, it's reasonable to say that protection against
> bad registration and unregistrations does matter -- a little -- and
> isn't critical.

As Greg said, "this is far more extreme corruption than should ever be
encountered in practice," so you should expect a much smaller effect from
occasional mistakes, and such an effect would be corrected with further
registrations anyway.  If you did have an extra false negative or two, you
would just correct them, and that'd be that.  I would go so far as to say
that protection against this doesn't really matter at all.  In fact, note
that you benefitted from occasional double-registrations.  If in doubt,
register again.  No harm.

Tom