CustomerInfoManager: post all unfinished transactions#2563
Conversation
7c3168c to
0a505fc
Compare
CustomerInfoManager: post all transactionsCustomerInfoManager: post all unfinished transactions
b517c2e to
64e8e38
Compare
There was a problem hiding this comment.
This is using the new async overload to simplify it.
cab519c to
1a48b0c
Compare
Codecov Report
@@ Coverage Diff @@
## main #2563 +/- ##
==========================================
- Coverage 88.02% 87.96% -0.06%
==========================================
Files 201 201
Lines 13997 13964 -33
==========================================
- Hits 12321 12284 -37
- Misses 1676 1680 +4
|
8e328bb to
188e971
Compare
tonidero
left a comment
There was a problem hiding this comment.
One comment about how to handle partial errors. Looks good otherwise
There was a problem hiding this comment.
Hmm I think this way we would override the previous result even if it was an error right? So if the last request succeeded but any of the previous ones failed, this would return success.
Should we post the first error if any? So the logic I think would be best would be:
- If any post fails, keep posting but keep track of the errors.
- Once all post end, if there were no errors, continue with the last success, or with the first error if any.
There was a problem hiding this comment.
Once all post end, if there were no errors, continue with the last success, or with the first error if any.
Considering that every post will post the entirety of the receipt, I think it would be best to be more optimistic with the results. This comes from calling getCustomerInfo. If we could post any of these, and get a CustomerInfo, it would be up-to-date.
I'll update the logic to pick any of the successes instead of "randomly" the last one.
There was a problem hiding this comment.
Ahhh right, it's different in iOS true. Then yeah, that makes sense 👍
There was a problem hiding this comment.
Done: f9ed157599c60f8912e3a4fa684270d33e9acd42
188e971 to
a76cdea
Compare
f9ed157 to
e47f3ab
Compare
**This is an automatic release.** ### New Features * `Offline Entitlements`: use offline-computed `CustomerInfo` when server is down (#2368) via NachoSoto (@NachoSoto) ### Bugfixes * `AppleReceipt.debugDescription`: don't pretty-print JSON (#2564) via NachoSoto (@NachoSoto) * `SK2StoreProduct`: fix crash on iOS 12 (#2565) via NachoSoto (@NachoSoto) * `GetCustomerInfo` posts receipts if there are pending transactions (#2533) via NachoSoto (@NachoSoto) ### Performance Improvements * `PurchasedProductsFetcher`: cache current entitlements (#2507) via NachoSoto (@NachoSoto) * Performance: new check to ensure serialization / deserialization is done from background thread (#2496) via NachoSoto (@NachoSoto) ### Dependency Updates * Bump fastlane from 2.212.2 to 2.213.0 (#2544) via dependabot[bot] (@dependabot[bot]) ### Other Changes * `CustomerInfoManager`: post all unfinished transactions (#2563) via NachoSoto (@NachoSoto) * `PostReceiptOperation`: added ability to also post `AdServices` token (#2566) via NachoSoto (@NachoSoto) * `Offline Entitlements`: improved computation log (#2562) via NachoSoto (@NachoSoto) * Added `TransactionPoster` tests (#2557) via NachoSoto (@NachoSoto) * Refactored `TransactionPoster`: removed 2 dependencies and abstracted parameters (#2542) via NachoSoto (@NachoSoto) * `CustomerInfoManagerTests`: wait for `getAndCacheCustomerInfo` to finish (#2555) via NachoSoto (@NachoSoto) * `StoreTransaction`: implemented `description` (#2556) via NachoSoto (@NachoSoto) * `Backend.ResponseHandler` is now `@Sendable` (#2541) via NachoSoto (@NachoSoto) * Extracted `TransactionPoster` from `PurchasesOrchestrator` (#2540) via NachoSoto (@NachoSoto) * `enableAdServicesAttributionTokenCollection`: added integration test (#2551) via NachoSoto (@NachoSoto) * `AttributionPoster`: replaced hardcoded strings with constants (#2548) via NachoSoto (@NachoSoto) * `DefaultDecodable`: moved to `Misc/Codable/DefaultDecodable.swift` (#2528) via NachoSoto (@NachoSoto) * `CircleCI`: specify device to run `backend_integration_tests` (#2547) via NachoSoto (@NachoSoto) * Created `StoreKit2TransactionFetcher` (#2539) via NachoSoto (@NachoSoto) * Fix load shedder integration tests (#2546) via Josh Holtz (@joshdholtz) * Fix doc on `Offering.getMetadataValue` (#2545) via Josh Holtz (@joshdholtz) * Extracted and tested `AsyncSequence.extractValues` (#2538) via NachoSoto (@NachoSoto) * `Offline Entitlements`: don't compute offline `CustomerInfo` when purchasing a consumable products (#2522) via NachoSoto (@NachoSoto) * `OfflineEntitlementsManager`: disable offline `CustomerInfo` in observer mode (#2520) via NachoSoto (@NachoSoto) * `BasePurchasesTests`: fixed leak detection (#2534) via NachoSoto (@NachoSoto) * `PurchaseTesterSwiftUI`: added `ProxyView` to `iOS` (#2531) via NachoSoto (@NachoSoto) * `PurchasedProductsFetcher`: removed `AppStore.sync` call (#2521) via NachoSoto (@NachoSoto) * `PurchaseTesterSwiftUI`: added new window on Mac to manage proxy (#2518) via NachoSoto (@NachoSoto) * `PurchasedProductsFetcher`: added log if fetching purchased products is slow (#2515) via NachoSoto (@NachoSoto) * `Offline Entitlements`: disable for custom entitlements mode (#2509) via NachoSoto (@NachoSoto) * `Offline Entitlements`: fixed iOS 12 tests (#2514) via NachoSoto (@NachoSoto) * `PurchasedProductsFetcher`: don't throw errors if purchased products were found (#2506) via NachoSoto (@NachoSoto) * `Offline Entitlements`: allow creating offline `CustomerInfo` with empty `ProductEntitlementMapping` (#2504) via NachoSoto (@NachoSoto) * `Offline Entitlements`: integration tests (#2501) via NachoSoto (@NachoSoto) * `CustomerInfoManager`: don't cache offline `CustomerInfo` (#2378) via NachoSoto (@NachoSoto) * `DangerousSettings`: debug-only `forceServerErrors` (#2486) via NachoSoto (@NachoSoto) * `CocoapodsInstallation`: fixed `Xcode 14.3.0` issue (#2489) via NachoSoto (@NachoSoto) * `CarthageInstallation`: removed workaround (#2488) via NachoSoto (@NachoSoto) --------- Co-authored-by: NachoSoto <ignaciosoto90@gmail.com>
…urchaseInDevicePostsReceipt` (#2589) Turns out this test was only passing because it was waiting for the receipt to be posted during a renewal, not the initial purchase! This updates the test to reflect the correct expectation (disabling renewals), and added the `XCTExpectFailure` so we can detect if it starts passing in a future iOS version. Also changed the test to not use `customerInfo()` because of #2563.
This expands the logic added in #2533 to post not only the first transaction, but all of them.
This is important for cases where there might be a consumable transaction. By handling all those, we ensure that we are able to process offline
CustomerInfolater on, instead of potentially failing due to a consumable transaction still unfinished.