> No, the boundary is intentionally the earlier one: I found that commit in GitHub, thanks for pointing it out. When I test locally *origin_in_the_future *case I get different results for positive and negative intervals (see queries #1 and #2 from above, they have same timestamp, origin and interval magnitude, difference is only in interval sign) - can it be that the version I downloaded from https://www.enterprisedb.com/postgresql-early-experience doesn't include commit with that improvement?
> I wonder if we should just disallow negative intervals here. I cannot imagine somebody using negative as a constant argument but users can pass another column as a first argument date or some function(ts) - not likely but possible. A line in docs about the leftmost point of interval as start of the bin could be helpful. Not related to negative interval - I created a PR for adding zero check for stride https://github.com/postgres/postgres/pull/67 and after getting it closed I stopped right there - 1 line check doesn't worth going through the patching process I'm not familiar with. >In the case of full units (1 minute, 1 hour, etc.), it gives the same result as the analogous date_trunc call, Was not obvious to me that we need to supply Monday origin to make date_bin(1 week, ts) produce same result with date_trunc Sorry for the verbose report and thanks for the nice function - I know it's not yet released, was just playing around with beta as I want to align CrateDB date_bin <https://github.com/crate/crate/issues/11310> with Postgresql On Thu, Jul 22, 2021 at 7:28 PM John Naylor <john.nay...@enterprisedb.com> wrote: > > On Thu, Jul 22, 2021 at 12:24 PM Bauyrzhan Sakhariyev < > baurzhansahar...@gmail.com> wrote: > > > > Is date_bin supposed to return the beginning of the bin? > > Thanks for testing! And yes. > > > And does the sign of an interval define the "direction" of the bin? > > No, the boundary is intentionally the earlier one: > > /* > * Make sure the returned timestamp is at the start of the bin, even if > * the origin is in the future. > */ > if (origin > timestamp && stride_usecs > 1) > tm_delta -= stride_usecs; > > I wonder if we should just disallow negative intervals here. > > -- > John Naylor > EDB: http://www.enterprisedb.com >
