I have encounter an unexpected result when running st_dumppoints() with
a partitioned dataset.
In this case the partitioned dataset has the following definition for
the master table partitioned by tc_date
CREATE TABLE maps.routelines (
tc_line_id serial4 NOT NULL,
tc_date date NOT NULL,
shape_id varchar(10) NOT NULL,
route_id varchar(10) NOT NULL,
route_short_name varchar(15) NULL,
route_long_name varchar(255) NULL,
route_desc varchar(50) NULL,
route_type int2 NULL,
direction_id int2 NOT NULL,
both_ways bool NULL,
geom public.geometry NOT NULL,
CONSTRAINT routelines_pk PRIMARY KEY (tc_date, shape_id)
);
The following query is successful.
select tc_date,shape_id,direction_id,route_id,(st_dumppoints(geom)).*
from maps.routelines
where route_type<=3
group by tc_date,shape_id,direction_id,route_id;
However, when I create a table from maps.routelines as shown below the
query fails with the message 'ERROR: column "routelines_not_school.geom"
must appear in the GROUP BY clause or be used in an aggregate function'
This is the process which results in the error.
drop table if exists routelines_not_school;
create temp table routelines_not_school as select * from maps.routelines
where route_type<=3; -- exclude school bus services
select tc_date,shape_id,direction_id,route_id,(st_dumppoints(geom)).*
from routelines_not_school
where route_type<=3
group by tc_date,shape_id,direction_id,route_id;
This is unexpected.
Any assistance appreciated.
Matt.
_______________________________________________
postgis-users mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/postgis-users
