Improved count_words_in_array(text[])

[peterlindsten]

Pure SQL version that should always perform faster

[peterlindsten]

5-6 times faster on my particular dataset.

Old:

Seq Scan on chunks_body_bm25i_docsbm25plus  (cost=0.00..4081.92 rows=9292 width=32) (actual time=0.086..5565.100 rows=9292 loops=1)
Planning Time: 0.051 ms
Execution Time: 5565.376 ms

New:

Seq Scan on chunks_body_bm25i_docsbm25plus  (cost=0.00..4081.92 rows=9292 width=32) (actual time=0.371..954.820 rows=9292 loops=1)
Planning Time: 0.147 ms
Execution Time: 955.082 ms