00:00:00

Count bad dataset rows once

The same row fails two rules and is charged twice against the allowed bad-row count. Trace individual rule results and overall acceptance. Count affected rows once without discarding the per-rule evidence. A teammate proposes the patch below. Review it against the full application contract, including unchanged code. Decide whether it is complete and safe to merge.

Review focus: Data Engineer

src/domain/validation.py

BeforeProposed

+2−2

3030

3131def accepted(results, maximum):

3232 maximum = integer(maximum, 'maxFailures')

33 failed_rows = []

33 failed_rows = set()

3434 for rule in results:

35 failed_rows.extend(rule['failures'])

35 failed_rows.update(rule['failures'])

3636 return len(failed_rows) <= maximum

Your review

0 / 10,000 characters · A verdict needs supporting evidence.

30 more characters to give a verdict