Support Swift bridging for RACSignalError using NS_ERROR_ENUM #150

sharplet · 2018-12-26T19:22:28Z

Redefines RACSignalError using NS_ERROR_ENUM, allowing it to bridge to Swift as a new RACSignalError struct containing static instances of RACSignalError.Code, and allowing it to be used in error pattern matching contexts.

sharplet · 2018-12-26T19:35:50Z

There’s a potential breaking change here: the symbols RACSignalErrorNoMatchingCase and RACSignalErrorTimedOut will no longer be available to Swift. I pushed commit 2334221, which adds them back in with a deprecated+renamed warning. Any thoughts on whether this is necessary? The downside is that it adds Swift code to a previously ObjC-only framework, which could cause build errors.

sharplet · 2018-12-26T19:47:54Z

Discoved RACCommandErrorDomain and RACSelectorSignalErrorDomain and added bridging for those as well.

sharplet · 2018-12-26T22:06:51Z

The pod lib lint failure appears to be a CocoaPods bug (CocoaPods/CocoaPods#8089):

 -> ReactiveObjC (3.1.0)
    - NOTE  | xcodebuild:  note: Using new build system
    - NOTE  | xcodebuild:  note: Planning build
    - NOTE  | xcodebuild:  note: Constructing build description
    - NOTE  | xcodebuild:  warning: Capabilities for App may not function correctly because its entitlements use a placeholder team ID. To resolve this, select a development team in the build settings editor. (in target 'App')
    - NOTE  | xcodebuild:  warning: Skipping code signing because the target does not have an Info.plist file. (in target 'App')
    - ERROR | [OSX] xcodebuild: Returned an unsuccessful exit code. You can use `--verbose` for more information.

Updating to CocoaPods 1.6.0.beta.2 locally resolved the error for me.

sharplet · 2018-12-26T23:07:48Z

For some reason, switching to Apple TV 4K instead of Apple TV 1080p worked, even though Apple TV 1080p clearly seemed to be in the list of available destinations.

The CocoaPods thing is still there, but I’m comfortable ignoring it until 1.6.0 is released.

mdiep · 2018-12-29T23:48:20Z

ReactiveObjC.xcodeproj/project.pbxproj

@@ -2579,13 +2600,19 @@
 					"DTRACE_PROBES_DISABLED=1",
 				);
 				INFOPLIST_FILE = ReactiveObjC/Info.plist;
+				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/../Frameworks @loader_path/Frameworks";


This shouldn't be needed, I don't think. It should already be specified in the xcconfig?

Right, this was probably an automatic Xcode change when adding the first Swift code to the target.

I removed all the redundant LD_RUNPATH_SEARCH_PATHS settings in commit 31e9f4a.

mdiep · 2018-12-29T23:56:09Z

ReactiveObjC/RACBlockTrampoline.m

@@ -28,7 +28,7 @@ - (instancetype)initWithBlock:(id)block {
 + (id)invokeBlock:(id)block withArguments:(RACTuple *)arguments {
 	NSCParameterAssert(block != NULL);

-	RACBlockTrampoline *trampoline = [[self alloc] initWithBlock:block];
+	RACBlockTrampoline *trampoline = [(RACBlockTrampoline *)[self alloc] initWithBlock:block];


Why was this needed?

While debugging the pod lib lint failures on CI I noticed warnings like this:

- WARN | [iOS] xcodebuild: /Users/travis/build/ReactiveCocoa/ReactiveObjC/ReactiveObjC/RACBlockTrampoline.m:31:49: warning: 'initWithBlock:' is only available on iOS 10.0 or newer [-Wunguarded-availability]

This is a false positive, because it’s confusing our -[RACBlockTrampoline initWithBlock:] selector for something like -[NSThread initWithBlock:], which has only been available since iOS 10. Adding the explicit cast shouldn’t be necessary (this seems like a compiler/analyser bug), but it does remove the warning.

mdiep · 2018-12-29T23:57:00Z

ReactiveObjC/Deprecations+Removals.swift

+public let RACSignalErrorNoMatchingCase = RACSignalError.noMatchingCase.rawValue
+
+@available(*, deprecated, renamed: "RACSignalError.timedOut")
+public let RACSignalErrorTimedOut = RACSignalError.timedOut.rawValue


This seems great. 👏

sharplet · 2018-12-30T16:18:33Z

ReactiveObjC.xcodeproj/project.pbxproj

 				PRODUCT_NAME = "$(PROJECT_NAME)Tests";
+				SWIFT_OPTIMIZATION_LEVEL = "-Onone";
+				SWIFT_VERSION = 4.2;


We could potentially lower this version to 4.0 or even 3.0 to keep support for compiling with older Xcode versions.

I'm not sure that's important.

mdiep · 2019-01-06T18:51:14Z

This looks good once the pod lib lint is fixed.

sharplet · 2019-01-07T15:25:05Z

@mdiep To fix it, we would either need to use a prerelease version of CocoaPods on CI or wait for 1.6.0 to land. With CocoaPods on a longer beta cycle, I’m not sure how long that will be.

Any preference?

mdiep · 2019-01-07T15:46:15Z

I'm fine with that. 👍🏻 Merge at your leisure!

sharplet · 2019-01-07T15:56:13Z

Oh weird, looks like I don’t have write access for ReactiveObjC?

Redefines `RACSignalError` using `NS_ERROR_ENUM`, allowing it to bridge to Swift as a new `RACSignalError` struct containing static instances of `RACSignalError.Code`, and allowing it to be used in error pattern matching contexts.

This allows us to use CocoaPods 1.6.0.beta, which fixes a bug in `pod lib lint` under Xcode 10.

mdiep · 2019-01-07T16:26:09Z

Oh weird, looks like I don’t have write access for ReactiveObjC?

Just invited you to the right team for that!

sharplet · 2019-01-07T20:58:25Z

@mdiep Thanks!

…dging Support Swift bridging for RACSignalError using NS_ERROR_ENUM

sharplet force-pushed the as-swift-error-bridging branch from 2145ce1 to 9014a17 Compare December 26, 2018 20:42

mdiep requested changes Dec 29, 2018

View reviewed changes

sharplet force-pushed the as-swift-error-bridging branch from d0dbcc8 to d8fb1f6 Compare December 30, 2018 16:12

sharplet commented Dec 30, 2018

View reviewed changes

mdiep approved these changes Jan 7, 2019

View reviewed changes

sharplet added 10 commits January 7, 2019 11:15

Update to Nimble 7.3.1

6787f5e

Add deprecations for error codes bridged to Swift

ffd5478

Add RACCommandError to bridged Swift errors

cbb7d90

Add RACSelectorSignalError to bridged Swift errors

6d4ae65

Use Xcode 10 on CI

e0770cd

Enable -Wstrict-prototypes

33a2bfe

Work around spurious warnings of -initWithBlock: availability

9a6dc6e

Switch to Apple TV 4K on CI

64b8527

Use prerelease version of CocoaPods on CI

7b150ab

This allows us to use CocoaPods 1.6.0.beta, which fixes a bug in `pod lib lint` under Xcode 10.

sharplet force-pushed the as-swift-error-bridging branch from 697cb5f to 7b150ab Compare January 7, 2019 16:15

sharplet merged commit a621018 into ReactiveCocoa:master Jan 7, 2019

sharplet deleted the as-swift-error-bridging branch January 7, 2019 20:58

haifengkao mentioned this pull request Jan 8, 2019

how to silence initWithBlock warning #152

Closed

stuartofmine pushed a commit to stuartofmine/ReactiveObjC_2023 that referenced this pull request Oct 18, 2023

Merge pull request ReactiveCocoa#150 from sharplet/as-swift-error-bri…

7d96881

…dging Support Swift bridging for RACSignalError using NS_ERROR_ENUM

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support Swift bridging for RACSignalError using NS_ERROR_ENUM #150

Support Swift bridging for RACSignalError using NS_ERROR_ENUM #150

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

mdiep Dec 29, 2018

sharplet Dec 30, 2018

sharplet Dec 30, 2018

mdiep Dec 29, 2018

sharplet Dec 30, 2018

mdiep Dec 29, 2018

sharplet Dec 30, 2018

mdiep Jan 6, 2019

mdiep commented Jan 6, 2019

sharplet commented Jan 7, 2019

mdiep commented Jan 7, 2019

sharplet commented Jan 7, 2019

mdiep commented Jan 7, 2019

sharplet commented Jan 7, 2019

Support Swift bridging for RACSignalError using NS_ERROR_ENUM #150

Support Swift bridging for RACSignalError using NS_ERROR_ENUM #150

Conversation

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

sharplet commented Dec 26, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mdiep commented Jan 6, 2019

sharplet commented Jan 7, 2019

mdiep commented Jan 7, 2019

sharplet commented Jan 7, 2019

mdiep commented Jan 7, 2019

sharplet commented Jan 7, 2019