v0.25.0

Minor Changes

• #285 6425ea9 Thanks @baseballyama! - Update postgresql-eslint-parser to 0.6.0. Reports now land on the statement
  or token they describe. Some existing reports move, so this is a minor
  release:

  • no-syntax-error shows PostgreSQL's own message for lexer errors, such as an
    unterminated string, instead of a JSON parse error.
  • A rule that reports on the second or a later statement in a file now
    reports at that statement's start, not at the end of the previous statement.
    This affects consistent-create-index-concurrently and no-grant-all, for
    example.
  • When a statement is preceded by a comment, rules such as
    require-table-columns report on the CREATE line, not the comment line.
  • eslint-disable-next-line placed above a statement now applies to it.
  • In the embedded-sql processor, statement-level reports on Prisma's
    $executeRaw / $queryRaw (for example require-where-in-delete) point at
    the start of the SQL statement. Reports on a node inside the statement do not
    move.
  • Rules that read tokens, such as no-identifier-too-long and
    consistent-as-for-column-alias, can report new problems. Like PostgreSQL,
    the parser now keeps these as single tokens: non-ASCII identifiers (名前),
    identifiers with $, prefixed strings (E'…'), and operators such as ->>.
    Review new reports after upgrading.

  The parser now also loads with @libpg-query/parser 17.8.0, and needs 17.6.6 or
  later.