release: the mint chain runs end to end against the real Stripe test API - #239
Merged
Merged
Conversation
…e test API (#238) `mintForProposal -> createExpressAccount -> recordConnectAccount -> the fields a registry entry carries` was unit- and mutation-tested and each Stripe call had been probed by hand, but the CHAIN had never executed in one piece anywhere: staging cannot rehearse it (`provisioningConfigured()` is `Boolean(PROVISION_GITHUB_TOKEN)` and that variable is absent there), so the first run of the hand-offs would have been a real restaurant's account. tests/e2e/connect-mint.spec.ts runs everything below `openProvisioningPr` against the REAL Stripe API on an `sk_test_` key and the suite's throwaway Postgres. Nothing is mocked (CLAUDE.md §7). It asserts the minted `acct_`, the locale-free `payments_link_url:` (and drives the middleware in French and in German to prove one stored URL serves both), the StripeConnectAccount row, what Stripe itself holds, that a cash-only tenant mints nothing, and that EUR is refused BEFORE the call — proven with a deliberately broken key whose 401 the CHF control does surface. The assertion the file exists for is idempotency, in both forms: a re-run reads our own row, and a run whose row is DELETED (the crash between the Stripe POST and the write) recovers the SAME account from the idempotency key rather than minting a twin. Stripe's own account list, filtered on `metadata[sofra_tenant]`, is the only oracle that can see that failure — the unique slug column cannot. Cleanup is asserted, not best-effort: every account is deleted (`deleted: true`, then GET -> 403) from an inventory taken from the database AND from Stripe, so an account created by a test that then failed is still removed. Guards mirror billing-mollie.spec.ts: the suite and CI refuse anything but `sk_test_`, and with no key the spec SKIPS with a stated reason. prisma/schema.prisma: `moduleFormat = "cjs"`. The generator was emitting an ESM-only client (`import.meta.url`) into a package that declares no `type`, so only a bundler could load `lib/db.ts` — which is why this chain had no way to be exercised outside a browser request. Output format only; lib/generated/ is gitignored and rebuilt on every build. Measured: the full unmocked suite is 83/83 green on this change (including the real Mollie first payment), and the Stripe test platform holds no `e2e-` account afterwards.
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Release — the mint chain runs end to end against the real Stripe test API
One commit (#238). Test coverage plus one output-format change.
The gap this closes.
mintForProposal -> createExpressAccount -> recordConnectAccount -> payments linkwas unit-tested and mutation-tested, and every Stripe call probed by hand, but the chain had never executed
in one piece in any environment —
sofra-stagingcannot rehearse it (noPROVISION_GITHUB_TOKEN). E8 wouldhave been its first run, on a real restaurant.
tests/e2e/connect-mint.spec.tsruns it against the REAL Stripe test API and the suite's throwaway Postgres.Green standalone and inside a full unmocked run (83/83, including the real Mollie first payment).
What it proves, measured rather than asserted
accept-language: franddeandgetting two different middleware redirects, not by inspecting a string.
crash window the table exists for) the next call recovers the same
acct_from the idempotency keyrather than minting a twin. Stripe's own list, filtered on
metadata[sofra_tenant], confirms one tenant oneaccount — which the unique slug column cannot see.
back with Stripe's 401 (instrument awake), EUR came back with the country refusal instead.
Cleanup inventories the database and Stripe, not what a test remembered, and asserts
deleted: truethenGET-> 403. Guards mirrorbilling-mollie.spec.ts: live keys hard-refused, and with no key the file skipswith a stated reason — CI shard 1 reported "4 skipped, 40 passed", so the skip path is exercised.
One production-adjacent change
prisma/schema.prismasetsmoduleFormat = "cjs". The generator was emitting an ESM-only client(
import.meta.url) into a package declaring notype, so only a bundler could loadlib/db.ts— the realreason this chain had never been exercisable outside a browser request. Output format only:
lib/generated/is gitignored and rebuilt on every build. Proven by the full 83/83 suite and by CI's build,prisma drift check and both Playwright shards.
Sonar: gate OK, 0 open issues; three
typescript:S1607transitioned to ACCEPTED with a rationale, the sametriage
billing-mollie.spec.ts:97already carries.