Re: [boost] 3-state boolean type

From: Alec Ross (alec_at_[hidden])
Date: 2002-08-02 02:23:49

...
>That's not the same as fuzzy logic, and it would be a very good use of the
>proposed 3-state data type, in my opinion.  The third state in this usage would
>be better named 'undecided' or 'undetermined', rather than 'indeterminate'.  
>For
>brevity, however, I suggest 'maybe' as the third state.  It's just as likely to
>be applicable as the other possibilities, and its meaning is clear as an
>alternative to 'true' or 'false'.
>

This - lazy evaluation and saving of the result - has been my main
experience too of the usage of 3-value logic.  But I have used it in
other ways too: notably using the "middle" value to mean "irrelevant".

I certainly could use a 3-state type.  My most common want would be for
the values to be settable at run-time, though they might be initialised
at compile-time.  (But, FWIW, when I've considered writing such a
facility I've balked at the exclusion of extending it to a tuple.  (Such
that the tuple could carry , separately, information on null-type,
empty-type, don't know, don't care ...)  This possible range of uses for
the 3-state type shows that there is indeed an issue about choosing the
name for the non-excluded middle state.
-- 
Alec Ross