Two scheduled workflows describe their own cron as daily when it is weekly:
.github/workflows/cron-vendor.yml:5: - cron: '0 10 * * 6' # everyday at 10am
.github/workflows/cron-conda.yml:8: - cron: '0 10 * * 6' # everyday at 10am
0 10 * * 6 is minute 0, hour 10, day-of-week 6 — Saturdays at 10:00 UTC
(no workflow sets a timezone: override, and scheduled workflows default to
UTC). The schedules themselves are correct and intended; only the comments are
wrong.
The repo's own prose already has it right, which is what makes the comments
misleading rather than merely terse: docs/source/contributing/workflows.rst:108
labels the node schedule (Saturday), releasing.rst:32 calls the vendor crawl
weekly, and tests/_dependency_gates.py:457 says "the three Saturday schedules".
Of the four other scheduled workflows, three carry no cron comment at all —
codeql-analysis.yml:10, deploy-pages.yml:16, python-compatibility.yml:9.
The fourth, lint.yml:138-139, already carries a correct one naming Saturday
and the sibling schedules. So the defect is confined to the two files above.
Surfaced by the cross-review on #963, which found the cron-vendor.yml instance;
the cron-conda.yml copy turned up on checking whether it was the only one.
Two scheduled workflows describe their own cron as daily when it is weekly:
0 10 * * 6is minute 0, hour 10, day-of-week 6 — Saturdays at 10:00 UTC(no workflow sets a
timezone:override, and scheduled workflows default toUTC). The schedules themselves are correct and intended; only the comments are
wrong.
The repo's own prose already has it right, which is what makes the comments
misleading rather than merely terse:
docs/source/contributing/workflows.rst:108labels the node
schedule (Saturday),releasing.rst:32calls the vendor crawlweekly, and
tests/_dependency_gates.py:457says "the three Saturday schedules".Of the four other scheduled workflows, three carry no cron comment at all —
codeql-analysis.yml:10,deploy-pages.yml:16,python-compatibility.yml:9.The fourth,
lint.yml:138-139, already carries a correct one naming Saturdayand the sibling schedules. So the defect is confined to the two files above.
Surfaced by the cross-review on #963, which found the
cron-vendor.ymlinstance;the
cron-conda.ymlcopy turned up on checking whether it was the only one.