standup
Standup — 2026-07-31
Busy day.
▶ Watch the 39s summary — Daily standup — Standup 2026-07-31
Daily standup for 2026-07-31. What shipped:
Busy day. Forty-plus PRs touched six different workstreams in the last 26 hours.
WishList connector reached completeness. #799–#892 together deliver transaction and subscription sync now that wlm_transactions/wlm_subscriptions exist. #890 fixed four review items that had never been committed: has_ledger() wasn’t escaping LIKE wildcards (so wp_wlm_transactions matched a leftover wpXwlm_transactions), and expires_at was reading only ExpiryDate—which reported lifetime-access for cancelled members. #927 fixed a 400-full-scan-per-sync problem where WishList’s COUNT queries (which can’t use an index) ran once per REST page; they’re now cached in a 10-minute transient at the dispatcher level, so the fix covers every module.
Multisite scoping shipped. #894 closes the email-join leak (#737), the BuddyPress wrong-table bug (#887), and LearnDash’s cross-subsite enrollment ids (#889). #933 handles the MemberMouse edge case separately: on legacy-DB installs MM_PREFIX is empty and the tables are genuinely network-wide with nothing to scope to, so the module now refuses rather than leaking.
WIF migration completed. #912–#925 walk the full §2a sequence: bring memberintel-deploy’s binding into Terraform, map repository_id on the provider, dual-bind all six SAs, broaden the condition, remove name-keyed grants, narrow to repository_id-only. Live-auth proof across deploy.yml, terraform-drift.yml, audit-runbook-check.yml, and plugin-release.yml confirms the gate is now rename-safe. Numeric id 1233497308 = caseproof/memberintel.
Other PRs worth noting:
- #904: achievement badges and a working nudge cadence—the reminder endpoint had never had a scheduler job, so it had never fired
- #920: GDPR/CCPA retention sweep + erase-by-email for anonymous contact submissions
- #882: onboarding follow-up now lands in a seeded conversation instead of an empty chat
- #863: lapsed trial classified correctly—not as a cancellation
- #821: over-budget retry timing corrected; a cancelled site loses a full cadence, not one tick
Notable. The nudge cadence fix in #904 found that the scheduled job had never existed—CRON_SECRET was never provisioned in infra/, so the endpoint fail-closed to 401 in every environment. Zero emails delivered since launch.
Window: 2026-07-30T13:00
→ 2026-07-31T15:23.025Z · Sources: memberintel @ a7ecd70, memberpress-intel @ dd7b305