Hi Karl,
You're correct that it is indeed the cartesian product in this case - it
produces one index entry for every unique tuple of values from the indexed
columns.
This gets slightly more complicated in the situation where the same column
is being indexed multiple times. In that situation, the
Thanks for the clarification, Nick.
Karl
On Mar 23, 5:58 am, Nick Johnson (Google) nick.john...@google.com
wrote:
Hi Karl,
You're correct that it is indeed the cartesian product in this case - it
produces one index entry for every unique tuple of values from the indexed
columns.
This gets
OK, I just watched Brett Slatkin's I/O talk [1] and he mentions cross
product a couple of times, so it seems that the use of the word
permutation in the docs is incorrect; the number of index entries is
indeed proportional to the Cartesian product, rather than
permutations which would lead to a