Pull request#2
Open
Gleb-Kovalenko wants to merge 256 commits into
Open
Conversation
2.1.0 restores the ability for TCA AppKit apps to access the `observe` function.
* Allow an alert to present another alert When we added support for vanilla SwiftUI modifiers, we lost the ability to present one alert after another because `nil` writes to the alert bindings unconditionally dismissed the feature, even if the feature was freshly presented. This fixes things by suppressing dismissal when the identity of a presented item has changed. Fix #3272. * wip
* `@preconcurrency @MainActor` isolation of `Store` * Remove unneeded `@MainActor`s * Remove thread checking code * Remove unneeded `@MainActor`s * Swift 5.10 compatibility fixes * wip * More 5.10 fixes * wip * fixes * wip * wip * up the timeout * wip * Fixes * Remove mainActorASAP in favor of mainActorNow. (#3288) * wip * Run swift-format * Update README.md * Fix integration tests. (#3294) * Fix integration tests. * wip * wip * Run swift-format * mainActorNow doesnt need escaping closure * wip * migration guide * wip * Update MigratingTo1.14.md --------- Co-authored-by: Brandon Williams <mbrandonw@hey.com> Co-authored-by: Brandon Williams <135203+mbrandonw@users.noreply.github.com> Co-authored-by: mbrandonw <mbrandonw@users.noreply.github.com>
* `@preconcurrency @MainActor` isolation of `Store` * Remove unneeded `@MainActor`s * Remove thread checking code * Remove unneeded `@MainActor`s * Swift 5.10 compatibility fixes * wip * More 5.10 fixes * wip * fixes * wip * wip * up the timeout * wip * Fixes * wip * wip * wip * wip * wip * wip * Main actor logger * wip * wip * wip * fix * fix * fix * fix compilation error * wip * wip * bring back yield --------- Co-authored-by: Brandon Williams <mbrandonw@hey.com>
Now that we've locked to Swift Navigation >=2.2, we get this method for free.
* fix Typo * delete deprecated content
* `@preconcurrency @MainActor` isolation of `Store` * Remove unneeded `@MainActor`s * Remove thread checking code * Remove unneeded `@MainActor`s * Swift 5.10 compatibility fixes * wip * More 5.10 fixes * wip * fixes * wip * wip * up the timeout * wip * Fixes * wip * wip * wip * wip * wip * Fix binding action sendability * Address more binding action sendability * more bindable action sendability * more bindable action warnings * fix --------- Co-authored-by: Brandon Williams <mbrandonw@hey.com>
* Require sendable IDs in cancellation/reducers Many of our APIs warn that we are passing hashable identifiers across concurrency boundaries, so we should require that they're sendable. * fix * fix * Fix
* `@preconcurrency @MainActor` isolation of `Store` * Remove unneeded `@MainActor`s * Remove thread checking code * Remove unneeded `@MainActor`s * Swift 5.10 compatibility fixes * wip * More 5.10 fixes * wip * fixes * wip * wip * up the timeout * wip * Fixes * wip * wip * wip * wip * wip * Fix binding action sendability * Address more binding action sendability * more bindable action sendability * more bindable action warnings * fix * Make `Effect.map` sendable * Make `Effect.actions` sendable Also moves it to the test target, since it's only really used there. * Make `AnyPublisher.create` sendable * Make `_SynthesizedConformance` sendable * Avoid non-sendable captures of `self` in reducers We can capture the sendable case path instead. * Make `ViewStore.yield` sendable * Address internal sendability warning * fix * Another small warning --------- Co-authored-by: Brandon Williams <mbrandonw@hey.com>
* Address effect cancellation sendability * fix * wip * wip
* Address `Reducer._printChanges()` sendability Because printing is done on a queue, both `State` and `Action` must be sendable. While `State` is easy enough to make sendable, it might be a pain to do so in a large, modularized application. Actions are not always so easy, but are in simple cases. Alternately, since this is a debugging affordance: 1. We could forego sendability since all we're doing is hitting it with a `Mirror` at the end of the day, and traffic the state/action along in a `nonisolated(unsafe)`. 2. We could ditch the queue...but that could affect the performance pretty negatively in some cases. * Unchecked send the debug printing
* Require main actor isolation in store collection * `preconditionIsolated` is not available in iOS <14 * Update Sources/ComposableArchitecture/Observation/IdentifiedArray+Observation.swift
…rray (#3346) * Added failing test to document behavior with observation and identified array. * Update Tests/ComposableArchitectureTests/ObservableTests.swift --------- Co-authored-by: Stephen Celis <stephen@stephencelis.com>
* Update 05-HigherOrderReducers CaseStudies * Replace alert(store:) to alert(_:) * View clean up. * Fixed alert. * Update DownloadComponent.swift --------- Co-authored-by: Brandon Williams <mbrandonw@hey.com> Co-authored-by: Stephen Celis <stephen.celis@gmail.com>
* Warn if bindable store binding action isn't processed Looks like the warnings we emit when we detect `BindingReducer` is missing are only applied to view stores, and were not ported over to the newer observable store bindings. This PR fixes that, though the main caveat is the messages can't seem to point to any good context. These bindings are derived from dynamic member lookup, which can't include source context like file/line. * wip * Add test
#3327) * Edit 04-Navigation-Lists-NavigateAndLoad file in SwiftUI CaseStudies * Update Examples/CaseStudies/SwiftUICaseStudies/04-Navigation-Lists-NavigateAndLoad.swift Co-authored-by: Stephen Celis <stephen.celis@gmail.com> * change selection Type to `Identified<Row.ID, Counter.State?>?` --------- Co-authored-by: Stephen Celis <stephen.celis@gmail.com> Co-authored-by: Stephen Celis <stephen@stephencelis.com>
* Trait-deprecate `Effect` for `EffectOf`. The shape of `Effect` is changing in 2.0 to be `Effect<State, Action>`, so to get ahead of the curve we are offering a trait to help folks migrate to `EffectOf<Feature>` (supported today) when explicitly referring to `Effect`. * wip * wip * Added: Streamlined `onChange` operator While the current version seems flexible its semantics are a little strange: it creates a reducer and throws it away on change. This is even incompatible with how 2.0 will do things, where mount/dismount events rapidly triggering would make reasoning about these reducers difficult. So instead let's prepare folks for 2.0 with a version that is more compatible, and really just streamlines things today. It is trait-deprecated so folks looking to migrate to 2.0 can opt-in and make changes as necessary. * Hard-deprecate `Store.withState` Missed this one with all the other `@ObservableState`-specific deprecations. Let's not worry about trait-based deprecation here. It should be a relatively unused API. * Trait-deprecate `StorePublisher` All other Combine interfaces have been hard-deprecated, for the most part, but it is possible that this change is viral (_e.g._ in a UIKit app), so I'm hiding it behind a trait for now. * Trait-deprecate `Effect.merge`, `Effect.concatenate`, `Effect.map` Since Swift concurrency landed, with `async`-`await` and task groups, we have pushed folks towards using these tools in a `.run` effect instead of `.merge` and `.concatenate`. While many still reach for these tools today, Composable Architecture 2.0 is making some fundamental changes to effects that make them more efficient and powerful but less transformable. To prepare folks, we should deprecate them behind our 2.0 trait. * Don't deprecate `merge` * Update Package.swift
* Trait-deprecate `Effect` for `EffectOf`. The shape of `Effect` is changing in 2.0 to be `Effect<State, Action>`, so to get ahead of the curve we are offering a trait to help folks migrate to `EffectOf<Feature>` (supported today) when explicitly referring to `Effect`. * wip * wip * Added: Streamlined `onChange` operator While the current version seems flexible its semantics are a little strange: it creates a reducer and throws it away on change. This is even incompatible with how 2.0 will do things, where mount/dismount events rapidly triggering would make reasoning about these reducers difficult. So instead let's prepare folks for 2.0 with a version that is more compatible, and really just streamlines things today. It is trait-deprecated so folks looking to migrate to 2.0 can opt-in and make changes as necessary. * Hard-deprecate `Store.withState` Missed this one with all the other `@ObservableState`-specific deprecations. Let's not worry about trait-based deprecation here. It should be a relatively unused API. * Trait-deprecate `StorePublisher` All other Combine interfaces have been hard-deprecated, for the most part, but it is possible that this change is viral (_e.g._ in a UIKit app), so I'm hiding it behind a trait for now. * Trait-deprecate `Effect.merge`, `Effect.concatenate`, `Effect.map` Since Swift concurrency landed, with `async`-`await` and task groups, we have pushed folks towards using these tools in a `.run` effect instead of `.merge` and `.concatenate`. While many still reach for these tools today, Composable Architecture 2.0 is making some fundamental changes to effects that make them more efficient and powerful but less transformable. To prepare folks, we should deprecate them behind our 2.0 trait. * Don't deprecate `merge` * Hard-deprecate `BindingViewState`/`BindingViewStore` Both of these concepts were technically obsoleted by 1.7. * wip * wip
* Trait-deprecate `Effect` for `EffectOf`. The shape of `Effect` is changing in 2.0 to be `Effect<State, Action>`, so to get ahead of the curve we are offering a trait to help folks migrate to `EffectOf<Feature>` (supported today) when explicitly referring to `Effect`. * wip * wip * Added: Streamlined `onChange` operator While the current version seems flexible its semantics are a little strange: it creates a reducer and throws it away on change. This is even incompatible with how 2.0 will do things, where mount/dismount events rapidly triggering would make reasoning about these reducers difficult. So instead let's prepare folks for 2.0 with a version that is more compatible, and really just streamlines things today. It is trait-deprecated so folks looking to migrate to 2.0 can opt-in and make changes as necessary. * Hard-deprecate `Store.withState` Missed this one with all the other `@ObservableState`-specific deprecations. Let's not worry about trait-based deprecation here. It should be a relatively unused API. * Trait-deprecate `StorePublisher` All other Combine interfaces have been hard-deprecated, for the most part, but it is possible that this change is viral (_e.g._ in a UIKit app), so I'm hiding it behind a trait for now. * Trait-deprecate `Effect.merge`, `Effect.concatenate`, `Effect.map` Since Swift concurrency landed, with `async`-`await` and task groups, we have pushed folks towards using these tools in a `.run` effect instead of `.merge` and `.concatenate`. While many still reach for these tools today, Composable Architecture 2.0 is making some fundamental changes to effects that make them more efficient and powerful but less transformable. To prepare folks, we should deprecate them behind our 2.0 trait. * Don't deprecate `merge` * Hard-deprecate `BindingViewState`/`BindingViewStore` Both of these concepts were technically obsoleted by 1.7. * Trait-deprecate `$store.scope(state: \.destination)` The key paths must align in 2.0, so for the migration to work it must be of the form `$store.scope(state: \.$destination)`.
* Trait-deprecate `Effect` for `EffectOf`. The shape of `Effect` is changing in 2.0 to be `Effect<State, Action>`, so to get ahead of the curve we are offering a trait to help folks migrate to `EffectOf<Feature>` (supported today) when explicitly referring to `Effect`. * wip * wip * Added: Streamlined `onChange` operator While the current version seems flexible its semantics are a little strange: it creates a reducer and throws it away on change. This is even incompatible with how 2.0 will do things, where mount/dismount events rapidly triggering would make reasoning about these reducers difficult. So instead let's prepare folks for 2.0 with a version that is more compatible, and really just streamlines things today. It is trait-deprecated so folks looking to migrate to 2.0 can opt-in and make changes as necessary. * Hard-deprecate `Store.withState` Missed this one with all the other `@ObservableState`-specific deprecations. Let's not worry about trait-based deprecation here. It should be a relatively unused API. * Trait-deprecate `StorePublisher` All other Combine interfaces have been hard-deprecated, for the most part, but it is possible that this change is viral (_e.g._ in a UIKit app), so I'm hiding it behind a trait for now. * Trait-deprecate `Effect.merge`, `Effect.concatenate`, `Effect.map` Since Swift concurrency landed, with `async`-`await` and task groups, we have pushed folks towards using these tools in a `.run` effect instead of `.merge` and `.concatenate`. While many still reach for these tools today, Composable Architecture 2.0 is making some fundamental changes to effects that make them more efficient and powerful but less transformable. To prepare folks, we should deprecate them behind our 2.0 trait. * Don't deprecate `merge` * Hard-deprecate `BindingViewState`/`BindingViewStore` Both of these concepts were technically obsoleted by 1.7. * Trait-deprecate `$store.scope(state: \.destination)` The key paths must align in 2.0, so for the migration to work it must be of the form `$store.scope(state: \.$destination)`. * TCA2 backport: enum scopes TCA2 is stricter about scoping along feature boundaries, which means composed scoping like `\.destination?.sheet` aren't desirable. But these APIs could be improved to begin with, and have in TCA2, so let's backport them to make the migration easier. ```swift // Before: $store.scope(state: \.$destination?.sheet, action: \.destination.sheet) // After: $store.scope(state: \.$destination, action: \.destination).sheet ```
* 1.25 Migration Guide * wip
Updated sharedHelper function signature for Effect transition.
While this compiles in Swift 6.3, it fails in Swift 6.2. This regression was introduced in #3858.
* Fix stack overflow related to new enum scopes Enum scopes are a new scoping overload introduced in #3882 to prepare folks for Composable Architecture 2.0. There was unfortunately a recursive call that the compiler didn't catch. This should fix it. * wip * wip * wip * wip
Enum scoping does not currently have a hard trait deprecation associated with it because overloads could cause existing apps to fail to compile with the dreaded "complex expression" error. Users should still have a migration path here, which is served by a trait that can be enabled temporarily.
We'll favor wrapping with `withAnimation` in TCA2.
Alerts and confirmation dialogs held in `@Reducer enum`s will require a bit of extra ceremony to be 2.0-compatible, but 2.0 will come with more flexible tooling around handling prompts.
…n raises deprecation warning (#3906) In a `View` context, reading a `@Shared` value from the store's state will raise a deprecation warning when the store's `Action` conforms to `BindableAction`. ```swift @Reducer struct Feature { @ObservableState struct State: Equatable { @shared(.inMemory("value")) var value: Int = 0 } enum Action: BindableAction { case binding(BindingAction<State>) } } struct FeatureView: View { let store: StoreOf<Feature> var body: some View { let _ = store.value // 🛑 Setter for 'value' is deprecated: Use '$shared.withLock' to modify a shared value with exclusive access; when constructing a SwiftUI binding, use 'Binding($shared)' } } ``` This is because the dynamicMemberLookup selects an overload that requires a `WritableKeyPath` which is enough to trigger Swift to require a setter for the property which is invalid for a `@Shared` value. Applying `@_disfavoredOverload` will drive the lookup toward an overload requiring a `KeyPath` instead. Co-authored-by: Sean <sean@snowfort.software>
Fixes a compiler error by helping it disambiguate between this implementation of ~= and the more constrained one declared in Binding+Observation.swift
* Fix enum scoping to alerts * wip
* Add `scope(_:action:)` and `Scope(_:action:)` To prepare for changes coming in 2.0, we are introducing new scoping overloads that omit the `state:` parameter label. This label is already omitted for some scoping operations (`ifLet`, `forEach`, etc.) and is just added noise for the dominant part of the operation. The old API has been trait-deprecated so that folks can migrate at their own pace without too much noise. * wip
* Update docs for Store. * wip
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
20 participants
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.
No description provided.