On Monday, October 18, 2021, Bryn Llewellyn <
bryn@yugabyte.com> wrote:
I can't find anything, neither on the page in question here on Range Types nor in the doc on the overlaps operator, about the semantics for when a duration collapses to an instant. Am I missing this too?
Same page:
- includes no points (and will be normalized to 'empty'
SELECT '[4,4)'::int4range;
The asymmetry stems from the definition of the same-valued half-open interval being normalized to an instant by the sql standard when dealing with overlaps but normalizing to empty for our range specification. Neither choice seems wrong on its face. It is what it is at this point - I defer to the commit message and mailing list history for any relevant “why” discussion.
David J.