I hit this question recently when searching for the details of the BitmapIndexScan implementation in PostgreSQL. What I liked a lot was a short and clear test case in the appendix, that is able to produce plans really similar to the initially posted one. This is quite a common need in my practice, especially for cases when I'm not allowed to share schema details in public. Would it be possible to explain how to come up with such a script?
In fact, I have one case for which I'm not able to produce a test data. It was posted to the pgsql-performance list a while ago: here asked Sep 12 '14 at 10:52 Victor Markus Winand ♦♦ |
I see. Well, that's really a trial and error effort. I'm sorry, but I don't have any recipie to follow which always yields test-data to reproduce some effect. answered Sep 14 '14 at 18:11 Markus Winand ♦♦ |