First, every type in postgres is user-defined, in the sense that its
binary structure and the arithmetic and comparison operations you can
perform on it are defined by a set of native C functions that are
present in the database executable or loaded as shared libraries.
Because of postgres's e
This is a very interesting option. My biggest concern is performance:
the project will require tables with millions of tuples. How does the
performance of such user created types compare to using native types? Or
are they 'built' using the same structure?
Thanks again!
Marc
Paul Tillotson wrote
Use a numeric type if you need more precision.
template1=# create domain BIGINT_UNSIGNED numeric(20,0) check (value >=
0 and value < '18446744073709551616'::numeric(20,0));
CREATE DOMAIN
template1=# create table foobar (i BIGINT_UNSIGNED);
CREATE TABLE
template1=# insert into foobar (i) values (-1
