[HACKERS] overlaps performance

[Grzegorz Jaśkiewicz]

Hey guys,




(c <= a AND d > a) OR ( c >= a AND c < b)

instead of

(a,b) overlaps (c,d)

any corner cases, or particular reasons ?

I am asking here, because I know there is bunch of people here that know the topic very well. I need to use few 'overlaps' for timedate in my query. And I guess it is quite inefficient there. So my question would be, why isn't postgresql using indexes for OVERLAPS, and why optimizer doesn't substitute it with something like: (source of example) http://www.depesz.com/index.php/2007/11/21/find-overlapping-time-ranges/

thanks.

signature.asc
Description: OpenPGP digital signature