Sufficool, Stanley wrote:
> How would the bounding box on a point be a problem with group by? I am
aware of the classic example of a line 0,0-5,5 and 0,5-5,0 being treated
as same when grouped. But does this really apply to points?
Unfortunately, yes. Consider these two close points that have the same
bounding box:
SELECT column1::box2d
FROM (VALUES
('POINT(0.123456789 0.123456789)'::geometry),
('POINT(0.123456788 0.123456788)'::geometry)) AS foo;
column1
------------------------------------------------------------------------------
BOX(0.123456783592701 0.123456783592701,0.123456791043282 0.123456791043282)
BOX(0.123456783592701 0.123456783592701,0.123456791043282 0.123456791043282)
(2 rows)
So, using a group by on the bounding box will treat these two points will group them together, just like your linear
example above.
If your dataset is of significant precision, grouping by the hex string of the
point works well.
Cheers,
Kevin
_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users
