should be it

external/duckdb/test/optimizer/perfect_ht.test

@@ -0,0 +1,56 @@
+# name: test/optimizer/perfect_ht.test
+# description: Test aggregates that can trigger a perfect HT
+# group: [optimizer]
+
+statement ok
+CREATE TABLE timeseries(year INTEGER, val INTEGER);
+
+statement ok
+INSERT INTO timeseries VALUES (1996, 10), (1997, 12), (1996, 20), (2001, 30), (NULL, 1), (1996, NULL);
+
+# this query uses a perfect aggregate HT
+query II
+EXPLAIN SELECT year, SUM(val), COUNT(val), COUNT(*) FROM timeseries GROUP BY year ORDER BY year;
+----
+physical_plan	<REGEX>:.*PERFECT_HASH_GROUP_BY.*
+
+statement ok
+PRAGMA perfect_ht_threshold=0;
+
+# if we set the threshold to 0, the perfect HT is not used anymore
+query II
+EXPLAIN SELECT year, SUM(val), COUNT(val), COUNT(*) FROM timeseries GROUP BY year ORDER BY year;
+----
+physical_plan	<!REGEX>:.*PERFECT_HASH_GROUP_BY.*
+
+statement ok
+PRAGMA perfect_ht_threshold=1;
+
+# if we set it too small, it is not used still
+query II
+EXPLAIN SELECT year, SUM(val), COUNT(val), COUNT(*) FROM timeseries GROUP BY year ORDER BY year;
+----
+physical_plan	<!REGEX>:.*PERFECT_HASH_GROUP_BY.*
+
+# we can also use it with many columns, as long as the threshold is high enough
+statement ok
+create table manycolumns as select i a, i b, i c, i d, i e, i f, i g, i h, i, i j from range(0,2) tbl(i);
+
+statement ok
+PRAGMA perfect_ht_threshold=30;
+
+query II
+explain select a, b, c, d, e, f, g, h, i, j FROM manycolumns GROUP BY 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
+----
+physical_plan	<REGEX>:.*PERFECT_HASH_GROUP_BY.*
+
+# the threshold has to be in range
+statement error
+PRAGMA perfect_ht_threshold=-1;
+----
+<REGEX>:.*out of range.*
+
+statement error
+PRAGMA perfect_ht_threshold=100;
+----
+<REGEX>:.*out of range.*