Fabien COELHO <coe...@cri.ensmp.fr> writes: >> I had in mind something more like the attached.
> Yep. > I'm not too happy that it mixes API levels, and about the int/double/int > path. > Attached an updated version which relies on pg_jrand48 instead. Hm, I'm not sure that's really an improvement, but I pushed it like that (and the other change along with it). > Also, as > the pseudo-random state is fully controlled, seeded test results are > deterministic so the expected value can be fully checked. I found that the "expected value" was different in v11 than HEAD, which surprised me. It looks like the reason is that HEAD sets up more/different RandomStates from the same seed than v11 did. Not sure if it's a good thing for this behavior to change across versions. regards, tom lane