postgresql/no-identifier-too-long

Why this matters

PostgreSQL stores identifiers in a fixed-width `name` column whose size is `NAMEDATALEN` (64 by default). Anything longer than `NAMEDATALEN - 1` bytes is **silently truncated** at parse time, so the object is created under a different name than written. Every later `DROP CONSTRAINT` / `ALTER ... RENAME` / `\d` that uses the original name then fails with `does not exist`. Bytes — not characters — count, so a 22-character CJK name is already over the limit.

Examples

CREATE TABLE this_is_a_very_very_very_very_very_very_very_very_long_table (id bigserial PRIMARY KEY);

ALTER TABLE items
  ADD CONSTRAINT items_code_check_that_is_unfortunately_far_too_long_for_postgresql CHECK (length(code) > 0);

CREATE INDEX an_index_name_that_exceeds_the_sixty_three_byte_postgres_limit ON items (code);

CREATE TABLE long_but_within_limit_table (id bigserial PRIMARY KEY);

ALTER TABLE items ADD CONSTRAINT items_code_non_empty CHECK (length(code) > 0);

CREATE INDEX items_code_idx ON items (code);

Configure it

// eslint.config.js
import postgresql from "eslint-plugin-postgresql";

export default [
  {
    files: ["**/*.sql"],
    languageOptions: {
      parser: postgresql.configs.recommended.languageOptions.parser,
    },
    plugins: { postgresql },
    rules: {
      "postgresql/no-identifier-too-long": [
        "error",
        {
          max: 63,
        },
      ],
    },
  },
];

Options

max integer default: 63

Maximum identifier byte length. Override only if your PostgreSQL build raises `NAMEDATALEN` from the default 64.