"David G. Johnston" <david.g.johnston@gmail.com> writes:
> On Wednesday, February 12, 2025, PG Bug reporting form <
> noreply@postgresql.org> wrote:
>> broadway=# select jsonb('[2,3,1]') - to_jsonb(1);
>> ERROR: operator does not exist: jsonb - jsonb
>> LINE 1: select jsonb('[2,3,1]') - to_jsonb(1);
> This would be a feature request, not a bug report.
Indeed. We have these cases today:
postgres=# \do -
List of operators
Schema | Name | Left arg type | Right arg type | Result type |
Description
------------+------+-----------------------------+-----------------------------+-----------------------------+-------------------------------------
...
pg_catalog | - | jsonb | integer | jsonb | delete
arrayelement
pg_catalog | - | jsonb | text | jsonb | delete
objectfield
pg_catalog | - | jsonb | text[] | jsonb | delete
objectfields
...
These are already less than consistent, for example this is allowed:
postgres=# select jsonb('[2,3,1]') - 'foo'::text;
?column?
-----------
[2, 3, 1]
(1 row)
but this not so much:
postgres=# select jsonb('{"a": 1, "b": 2}') - 1;
ERROR: cannot delete from object using integer index
To invent jsonb - jsonb, you'd need to define its behavior for
every possible JSON structure on each side. I foresee plenty of
bikeshedding.
regards, tom lane
