Stamatis Zampetakis created CALCITE-6704:
--------------------------------------------
Summary: Add limit parameter to restrict the result size of
UniqueKeys metadata
Key: CALCITE-6704
URL: https://issues.apache.org/jira/browse/CALCITE-6704
Project: Calcite
Issue Type: Improvement
Components: core
Affects Versions: 1.38.0
Reporter: Stamatis Zampetakis
Assignee: Stamatis Zampetakis
In some cases, the number of unique keys grows exponentially for certain
relational expressions.
Consider a table with a composite key \{k1, k2, k3\} and the following query:
{code:sql}
SELECT k1, k1, k2, k2, k3, k3 FROM composite
{code}
The number of minimal unique keys for this query is 2^3:
{noformat}
{0, 2, 4}, {0, 3, 4}, {1, 2, 4}, {1, 3, 4}, {0, 2, 5}, {0, 3, 5}, {1, 2, 5},
{1, 3, 5}
{noformat}
In this case, the size of *minimal keys* is exponential to the number of
columns in the composite key. The number of keys is f(x,y) = x^y where _x_ is
the number of repetitions of each column and _y_ is the number of columns in
the composite key. The example is taken out from CALCITE-6640.
A large number of unique keys may occur in other use-cases as well. Computing
the entire result set may become prohibitively expensive and cause OOM and
other failures.
To prevent this from happening we propose to add a new parameter to the
UniqueKeys interface allowing the users to specify a limit on the size of the
result set.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
