On Sat, 19 Feb 2005, peter royal wrote:
> i'm mainly concerned about single-column functional indexes right now.
> being able to get upper(a) as the column name would be enough now. but
> tom's method for working with multi-column functional indices would be
> a bonus too.
>
> would a patch to have DMD.getIndexInfo() return this information be
> accepted into the tree? if so, i'll work one up.
Yes, a column name of "pg_expression_N" is rather useless. upper(a) may
not be a column name, but it is more accurate and informative.
Kris Jurka