Data Type Best Practices

pgfence catches common type choices that will cause problems down the line.

#	Pattern	Risk	Suggestion
20	ADD COLUMN ... json	LOW	Use jsonb: json has no equality operator and is generally slower
21	ADD COLUMN ... serial	LOW	Use GENERATED BY DEFAULT AS IDENTITY: SERIAL creates implicit sequences with ownership quirks and permission issues
22	integer / int columns	LOW	Use bigint: changing to bigint later requires an ACCESS EXCLUSIVE rewrite
23	varchar(N) columns	LOW	Use text: increasing varchar length requires ACCESS EXCLUSIVE on older Postgres versions
24	timestamp without time zone	LOW	Use timestamptz to avoid timezone-related bugs
25	char(N) / character(N) columns	LOW	Use text: char(n) pads values with spaces, wastes storage, and changing the length later requires ACCESS EXCLUSIVE + table rewrite
26	serial / bigserial / smallserial columns	LOW	Use GENERATED BY DEFAULT AS IDENTITY: IDENTITY columns have cleaner ownership semantics, avoid implicit sequence permission issues, and are the SQL standard