Answer by Schwern for Grouping a table by its type on different columns

SQL queries are row based, but you want columns. This requires a pivot. We need something in common for each row so we can group them together. We can add a row_number to each row to put them in order. Then we pivot them into individual columns by grouping by the row number.

with ordered_things as (  select    row_number() over(partition by type order by id asc) as rn,    id, type  from things)select  coalesce(max(id) filter(where type = 'ACCOUNT'), '') as account_id,  coalesce(max(id) filter(where type = 'CLIENT'), '') as client_id,  coalesce(max(id) filter(where type = 'ORDER'), '') as order_idfrom ordered_thingsgroup by rnorder by rn

Since we know there's only one ID for each type in the group, max(id) is just a way to satisfy the need for each column of a group by query to be an aggregated function.

coalesce just gets rid of unsightly nulls when there isn't an equal number of ids for each type.

Demonstration.