This is a query pattern for which I found the answer many years ago:
SELECT *
FROM mytable a
JOIN mytable b on a.starttime <= b.endtime
and a.endtime >= b.starttime
and a.name != b.name; -- ideally, this would compare a "key" column, eg id
To find "any overlap", you compare the opposite ends of the timeframe with each other. It's something I had to get a pen and paper out for and draw adjacent ranges to realise that the edge cases boiled down to this comparison.
If you want to prevent any rows from overlapping, put a variant of this query in a trigger:
create trigger mytable_no_overlap
before insert on mytable
for each row
begin
if exists (select * from mytable
where starttime <= new.endtime
and endtime >= new.starttime) then
signal sqlstate '45000' SET MESSAGE_TEXT = 'Overlaps with existing data';
end if;
end;
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…