It seems intuitively correct that if you have a large table having a highly-skewed, indexed, histogram-analyzed column of values "A" and "B", like ...
.. then Oracle would use the index for predicates like ...Code:select my_col,count(*)
from my_table
group by my_col;
MY_COL COUNT(*)
------ --------
A 99999999
B 1
MY_COL='B'
... or ...
MY_COL!='A'
... or ...
MY_COL='C'
... and would use FTS for predicates like ...
MY_COL='A'
... or ...
MY_COL!='B'
Unfortunately, I haven't been able to put together a test case to demonstrate it, so maybe theory don't apply here.