Changelog

All notable changes to Generous.nz are recorded here, grouped by version. Every deployed build has a version number, a git tag, and an entry below.

Versioning scheme

Versions follow MAJOR.MINOR.PATCH:

0.x — pre-launch. Minor bumps mark a completed build phase or major feature set; patches are bug fixes on the current build.
1.0.0 — first public launch.
1.x+ — post-launch: minor = new features, patch = fixes, major = breaking changes or significant re-architecture.

Each version is tagged in git (v0.2.0) and the current version is kept in package.json. See docs/version-control.md for how to cut a release.

Referral attribution is now first-touch for both fundraiser and business links: the first invite a visitor opens is saved for 5 days and later links never override it, so the first referrer keeps the bonus. Landing pages show the credited inviter when an invite is already saved.

Fundraiser referral landing page upgraded to match the business page: a prominent "Your invite bonus" card showing $100 of first fundraising fee-free ($0.00 platform fee on the first $100, then 5% + GST only when you raise, donors can cover the fee) plus a trust strip (direct Stripe payments, AI fraud checks, made in NZ, no monthly fees).

Referral links (/r/G-…) now open a tailored sales landing page instead of redirecting straight to sign-up. Business links pitch the first $1,000 in sales fee-free, direct payments and order fulfilment; fundraiser links pitch the first $100 raised fee-free with the full range of raise types and AI tools. Each page shows the inviter, the offer, how it works, FAQ, and a clear CTA, and the 5-day referral cookie is still set on load (via a server action) so the reward is claimed at sign-up. Unknown/invalid links show a friendly generic page instead of an error.

Purchase history now survives item deletion. Verified at the database level: Order → Product/Fundraiser and Donation → Fundraiser use ON DELETE RESTRICT, so hard-deleting a product or fundraiser with orders fails instead of wiping them, and the app's own remove/delete flows were already soft (status REMOVED/ARCHIVED). The one cascade that could erase sold vouchers — VoucherCode.wholesaleProductId → CASCADE — is now SET NULL, so purchasers keep their vouchers in "My purchases" even if a wholesale product is hard-deleted. Wholesaler voucher lists/export handle removed products gracefully.

docs/bulk-test-runbook.md updated: disputes/refunds/payouts on seller accounts are now covered by the daily reconciliation job (v0.9.187); only recurring-donation renewals still need the seller webhook secret.

Daily connected-account reconciliation (lib/payments/stripe-sync.ts, wired into /api/cron/reconcile which the daily app cron already calls). With the direct-charge model, Stripe sends dispute, refund and payout events to the seller's own account rather than the platform webhook — this job now polls every connected account over a rolling lookback (default 7 days, RECONCILE_LOOKBACK_DAYS) and replays the same logic the webhook would have run: mark orders/donations disputed or refunded, hold/cancel/restore commission, apply fee credits, notify sellers, raise moderation flags, and keep payout history current. Everything is idempotent, so missed webhooks self-heal on the next run, and seller-initiated refunds (which the webhook never saw) are now captured too.

docs/bulk-test-runbook.md corrected after the second full test run: Step 3 now makes clear that create-checkouts-2.js is only for re-creating wholesale checkouts that failed in the main script (running it otherwise duplicates orders), and stale checkouts2.json should be deleted before a fresh run. .attach-shared-account.js (server helper, not app code) was also fixed to recreate a plain index so re-runs work.

docs/bulk-test-runbook.md: the full automated test runbook for Stripe test mode. Saying "run testing" in a future conversation executes Steps 0–6 (test-mode check, seed data, shared test account attach, checkout creation, automated card payments, database/email verification, cleanup) and reports results per scenario.