[LucaCappelletti94]

ALTER DEFAULT PRIVILEGES was not a recognised ALTER target, so it failed at the DEFAULT keyword.

The statement now parses into AlterDefaultPrivileges, carrying the optional FOR ROLE and IN SCHEMA lists and the abbreviated GRANT or REVOKE, so it round trips rather than being consumed and discarded.

I tried to maximize code reuse from analogous structs.
All of these PRs come from trying to parse pg_dump -s outputs.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.62%. Comparing base (424378c) to head (d59af73).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2417      +/-   ##
==========================================
+ Coverage   81.52%   81.62%   +0.09%     
==========================================
  Files          42       42              
  Lines       34205    34368     +163     
  Branches    34205    34368     +163     
==========================================
+ Hits        27885    28052     +167     
  Misses       2770     2770              
+ Partials     3550     3546       -4     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sabir-akhadov-localstack]

Two non-blocking comments inline.

[sabir-akhadov-localstack]

The current docs also list ON LARGE OBJECTS (PG 17+), so ALTER DEFAULT PRIVILEGES GRANT SELECT, UPDATE ON LARGE OBJECTS TO app fails here; a LargeObjects variant matched via parse_keywords(&[LARGE, OBJECTS]) would cover it.

[LucaCappelletti94]

Done

[sabir-akhadov-localstack]

DefACLOptionList in gram.y is an unordered list, so ALTER DEFAULT PRIVILEGES IN SCHEMA s FOR ROLE r GRANT SELECT ON TABLES TO app is valid PG but fails here with Expected: GRANT or REVOKE; looping over the two options until neither matches would accept it. pg_dump always emits FOR ROLE first, so non-blocking.

[LucaCappelletti94]

FOR ROLE and IN SCHEMA are now one ordered option list, so either order parses and prints as written, and repeatitions are rejected, as PostgreSQL does.