I'd like to know what the motivation behind storing the numeric amount in postgres composite types as a numeric(20,8) is. With this setup, storing "USD 10.00" for example, turns into "USD 10.00000000" in the db, and then when retrieving it, it retains the precision.

I wanted more control over how much precision is stored for certain columns. I changed the composite money type from what the readme mentions:

CREATE TYPE public.money_with_currency AS (currency_code char(3), amount numeric(20,8))

To this:

CREATE TYPE public.money_with_currency_new AS (currency char(3), amount numeric)

It works much better for my use cases. Is there any reason why this shouldn't be the default?