Re: Collation order for btree-indexable datatypes

Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Comparing NaN/Invalid seems so off the beaten path that we would just
> wait for 7.2.  That and no one has reported a problem with it so far.

Do you consider "vacuum analyze" on the regression database to be
off the beaten path?  How about creating an index on a numeric column
that contains NaNs, or a timestamp column that contains Invalid?

Unless you believe these values are not being used in the field at all,
there's a problem.  (And if you do believe that, you shouldn't be
worried about my changing their behavior ;-))
        regards, tom lane