### Context:
`#if os(iOS)` no longer evaluates to `true` when compiling for
`visionOS`
Which means we need `|| os(xrOS)`.
But to support older Xcode we need `#if swift(>=5.9) && os(xrOS)`.
But, fun fact, short-circuiting doesn’t work in `Xcode 13`.
And we still need that to run iOS <15 tests (`FB12223404` /
`FB12805891`)
💩
This works around that by defining our own `VISION_OS` flag.
### Changes:
- Added `VISION_OS` compile time flag to determine SDK
- Replaced `#if os(xrOS)` with `#if VISION_OS`
- Added flag to `Package.swift` (thanks @neonichu for helping with this)
Created `Async` with some shared methods that make this conversion
simpler in a type-safe way.
Helps turn this:
```swift
_ = Task<Void, Never> {
do {
completion(.success(try await f()))
} catch {
completion(.failure(ErrorUtils.error())
}
}
```
Into this:
```swift
Async.call(with: completion) {
try await f()
}
```
This is the last change from what was started with #1437. That PR introduced a few private error types, but for the errors returned in the public `APIs` we were still passing around `Error`s everywhere, which didn't provide any compile-time guarantees.
### Goals:
- Ensure that public APIs *only* return `ErrorCode`s wrapped with the additional `userInfo`, and not accidentally return other types like `SKError`, `BackendError`, `ErrorCode` directly, etc.
- Guarantee type-safety when passing errors around within our internal implementations instead of assuming certain thrown errors are of a particular type.
### Changes:
- Created `PurchasesError` for wrapping an `ErrorCode` with the additional `userInfo`.
- Changed returned errors from `Error` to `NSError`. This is a change in the API, but the new type is a more specific `Error` so it shouldn't lead to any changes in client apps.
- Because `PurchasesError` isn't implicitly convertible to `NSError`, this ensures that we don't return those values directly, and instead use `PurchasesError.asPublicError`.
- Changed `ErrorCodeConvertible` to a now more precise `PurchasesErrorConvertible`.
Thanks to this new type-safety this PR fixes at least 3 bugs where we were returning private errors instead of `ErrorCode`s.
### Testing:
Thanks to #1871 we know that the returned errors are still convertible to `ErrorCode` so users can `switch` over them.
Fixes [CF-19].
This allows specifying `CustomerInfoManager` fetching behavior, both as a public API for users, and also internally.
Some future uses of `CustomerInfoManager` might want to ensure that we don't rely on cached data, for example.
Thanks to the `APITester` we know that the existing methods still work, this is purely an additive change.
This is the core of the change:
```swift
/// Specifies the behavior for a caching API.
@objc(RCCacheFetchPolicy)
public enum CacheFetchPolicy: Int {
/// Returns values from the cache, or throws an error if not available.
case fromCacheOnly
/// Always fetch the most up-to-date data.
case fetchCurrent
/// Returns the cached data if available and not stale, or fetches up-to-date data.
case notStaleCachedOrFetched
/// Default behavior: returns the cached data if available (even if stale), or fetches up-to-date data.
case cachedOrFetched
/// Default ``CacheFetchPolicy`` behavior.
public static let `default`: Self = .cachedOrFetched
}
```
The new `CustomerInfoManagerGetCustomerInfoTests` check all the different combinations of fetching behavior.
I've taken some of the existing tests and refactored them to be `async`, as well as added a bunch more tests.
> **Note**:
There is also one big hidden bug fix:
`Purchases.getCustomerInfo` now calls `BackendError.asPurchasesError`, so we don't end up returning an internal error type.
This was missed because we always return generic `Error` types, so we have no way to know at compile time that we only return `ErrorCode`s.
[CF-19]: https://revenuecats.atlassian.net/browse/CF-19?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ
Fixes [CF-684] and #1594.
`SystemInfo.appleSubscriptionsURL` was a short URL that didn't pass the `SystemInfo.isAppleSubscription(managementURL:)` test. This is fixed and covered by a test now.
### Other changes:
- Made `SystemInfo.isAppleSubscription(managementURL:)` and `SystemInfo.appleSubscriptionsURL` static since they don't need any properties in `SystemInfo`.
- Forced-unwrapped `SystemInfo.appleSubscriptionsURL` to remove unnecessary error handling. This is safe because the URL is provided statically at compile time. Per the docs "This initializer returns nil if the string doesn’t represent a valid URL. For example, an empty string or one containing characters that are illegal in a URL produces nil.". This is known at compile time, and covered by tests, so it could never return `nil`.
- Combined redundant tests in `ManageSubscriptionsHelper`. They were all doing exactly the same for both platforms.
[CF-684]: https://revenuecats.atlassian.net/browse/CF-684?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ
A lot of our classes depended on the mutable `IdentityManager`, when in reality they only needed a read-only `currentAppUserID`.
This PR changes the requirement, with the following advantages:
- Ensure all these classes (`AttributionPoster`, `PurchasesOrchestrator`, `TrialOrIntroPriceEligibiltyChecker`, `BeginRefundRequestHelper`, `ManageSubscriptionsHelper`) cannot logIn / logOut, only read the current user
- Simplify tests by only requiring a smaller type
- Simplifying instantiation of these types outside the normal `Purchases` initialization.
I've also created `MockCurrentUserProvider`, but changed the existing `MockIdentityManager` so that calls to `logIn`/`logOut` just fail. These aren't mocked, and they wouldn't have an impact on the mocked readable values, so calling them is most likely unexpected and should make the test fail.
NachoSoto