[nikolauspschuetz]

7 is a standard alias for Sunday (0), and croniter accepts it in the classic 5-field form. value_alias skipped the 7 -> 0 conversion in the 6-field (seconds) and 7-field (year) forms, so 7 was rejected there as out of range:

>>> croniter.is_valid("0 0 * * 7")      # 5-field
True
>>> croniter.is_valid("0 0 * * 7 0")    # 6-field
False
>>> croniter.is_valid("0 0 * * 7 0 *")  # 7-field
False

The guard is there to keep rejecting 0 as a day/month in those forms; day-of-week was swept into it. Dropping day-of-week from the two branches makes 7 -> Sunday work in every form while leaving the month/day 0 rejection unchanged. Regression test added; full suite green.

[potiuk]

LGTM — correct fix at the right layer. Dropping DOW_FIELD from the 6- and 7-field skip lists in value_alias lets 7 alias to Sunday like it does in the 5-field form. After the change, 5-7, 0-7, 7#2, L7, 7/2 and second_at_beginning all behave like 5-field. A side effect is that H(0-7) no longer fails for some hash ids. The new test fails on main, and CI is green.

• Heads-up: #281 edits the same line, adding DAY_FIELD for the 6-field form (#280). Whichever lands second needs a rebase. The end state should be [DAY_FIELD, MONTH_FIELD] for every form.
• nit (optional): the test only covers a bare 7. One range (5-7) and one #/L case would also exercise the low, high = value_alias(...) and nth paths.

---

This review was drafted by an AI-assisted tool and
confirmed by an croniter maintainer. The maintainer
approving this PR has read the findings and signed off. If
something feels off, please reply on the PR and a maintainer
will follow up.