Is your proposal related to a problem?

I am able to filter tables by computed fields and select computed field columns in queries, but I cannot aggregate (e.g. sum) by computed fields.

Here is a repro Hasura Cloud link: https://cloud.hasura.io/project/de479b35-64a2-4211-838b-cf5a4b49ee1a/details

Here is a short video explaining what I mean: https://www.loom.com/share/df97ba8eb905449c9807e433a77e7aab

this is the PostgreSQL schema:

CREATE TABLE code (
    id uuid DEFAULT gen_random_uuid() PRIMARY KEY,
    name text NOT NULL,
    negate boolean NOT NULL DEFAULT false
);

CREATE UNIQUE INDEX code_pkey ON code(id uuid_ops);

CREATE TABLE payroll_entry (
    id uuid DEFAULT gen_random_uuid() PRIMARY KEY,
    code_id uuid NOT NULL,
    source_amount numeric NOT NULL
);

CREATE UNIQUE INDEX payroll_entry_pkey ON payroll_entry(id uuid_ops);

CREATE OR REPLACE FUNCTION public.get_payroll_entry_amount(entry payroll_entry)
 RETURNS numeric
 LANGUAGE sql
 STABLE
AS $function$
  select (case when negate then -1 else 1 end) * entry.source_amount from code where entry.code_id = code.id
$function$
;

The practical issue for me is that I have a paginated, filterable table with tens of thousands of rows, and I don't want to have to load it all up into memory to sum up the values when the database can do that just fine

Describe the solution you'd like

expose computed fields in aggregations on tables

Describe alternatives you've considered

Not really anything viable I think.