Open Source Platform for DTC, B2B Commerce, Marketplaces & Omnichannel. REST APIs, TypeScript SDKs, and production-ready Next.js storefront. Self-host it. Own your data. No vendor lock-in. Zero platform fees.
Ruby b2b-commercee-commerceecommerceecommerce-apiecommerce-frameworkecommerce-platformheadlessheadless-commerceheadless-ecommercemarketplacemulti-tenantmulti-vendormulti-vendor-ecommerceopen-sourcespree-commerce
Correct comments that still described a timeout as a refusal
Damian Legawiec5b045f291c69
diff --git a/spree/core/app/workflows/spree/seller_payouts/sweep.rb b/spree/core/app/workflows/spree/seller_payouts/sweep.rbindex d5e02d3fb5..32a1227024 100644--- a/spree/core/app/workflows/spree/seller_payouts/sweep.rb+++ b/spree/core/app/workflows/spree/seller_payouts/sweep.rb@@ -12,10 +12,14 @@ module Spree# transfers with no payout — so a batch can never be swept twice, and a# re-run after a crash finds nothing left to take.#- # A failed send releases them again, because a payout is created here and+ # A refused send releases them again, because a payout is created here and# nowhere else: transfers left stamped to a failed one would be invisible# to every later sweep while the balance still reported them owed. The# next sweep is the retry.+ #+ # A send whose outcome nobody knows is the exception, and keeps them. The+ # money may already have gone, so releasing them is how the same earnings+ # get sent twice — they wait with the settlement until it resolves.class Sweep < Spree::Workflowhooks :validate, :after_sweepdiff --git a/spree/providers/stripe/app/models/spree_stripe/payout_provider.rb b/spree/providers/stripe/app/models/spree_stripe/payout_provider.rbindex 7197f79356..1d2cf31633 100644--- a/spree/providers/stripe/app/models/spree_stripe/payout_provider.rb+++ b/spree/providers/stripe/app/models/spree_stripe/payout_provider.rb@@ -16,8 +16,11 @@ module SpreeStripe# is going to do, and a webhook confirms when it did.## Every call carries an idempotency key derived from the ledger row, so a- # retry after a timeout finds the movement it already made instead of making- # a second.+ # retry finds the movement it already made instead of making a second. That+ # holds only while Stripe keeps the record, which is a day — so a call whose+ # answer was lost is reported as ambiguous rather than failed, and core+ # holds the settlement instead of retrying it later on a key that has since+ # expired.class PayoutProvider < Spree::PayoutProvider::Base# Stripe is looking at the account, and there is nothing the seller can do# about it — different words and no button, rather than an invitation to@@ -221,11 +224,16 @@ module SpreeStripe# Keyed to the earnings being settled, not to the settlement row.#- # A payout that times out is marked failed and releases its earnings, so- # the next sweep assembles a *new* row for the same money — and a key- # derived from that row would be a new key, which is Stripe's cue to send- # the money a second time. The transfers are what the two attempts have in- # common, so they are what identifies the operation.+ # A refused payout releases its earnings, so the next sweep assembles a+ # *new* row for the same money — and a key derived from that row would be+ # a new key, which is Stripe's cue to send the money a second time. The+ # transfers are what the two attempts have in common, so they are what+ # identifies the operation.+ #+ # This protects a retry only while Stripe keeps the record, which is a+ # day. It is why a payout whose outcome was never established is held+ # rather than retried at all: past that window the key is no defence, so+ # core keeps those earnings claimed instead of relying on one.def payout_idempotency_key(seller_payout)transfer_ids = seller_payout.transfers.order(:id).pluck(:id)digest = Digest::SHA256.hexdigest(