QUALIFY filters window results inline (dialect); GROUP BY ALL and friends are conveniences, not concepts.

Each sugar names a ritual: QUALIFY = the subquery-then-filter of top_n_per_group; GROUP BY ALL = 'group by everything I didn't aggregate' (with the caution that an added SELECT column silently changes the grouping); EXCLUDE = 'all but these'. The portability tax is the second half: none is universal — code that ships to Postgres expands QUALIFY back to the subquery, ALL back to the explicit list, EXCLUDE back to named columns.

SELECT
  cart,
  twist
FROM
  carts
QUALIFY
  ROW_NUMBER() OVER (
        PARTITION BY
          cart
        ORDER BY
          n DESC
      ) = 1
ORDER BY
  cart;

SELECT
  cart,
  twist
FROM
  (
    SELECT
      cart,
      twist,
      ROW_NUMBER() OVER (
        PARTITION BY
          cart
        ORDER BY
          n DESC
      ) AS rn
    FROM
      carts
  )
WHERE
  rn = 1
ORDER BY
  cart;

Both return the best twist per cart, but QUALIFY is engine sugar — it filters the window inline where DuckDB/Snowflake/BigQuery allow it, and code that ships to Postgres must expand it back to the subquery-then-filter.

Use them where they live; keep the portable expansion next to anything that travels.

QUALIFY / GROUP BY ALL / SELECT * EXCLUDE in DuckDB/Snowflake/BigQuery; Postgres needs the explicit expansions.