XFAIL GRDB and ObjectMapper

[rudkx]

XFAIL GRDB
swiftlang/swift#13342 uses conditional conformance to make Range a Collection when its Bound is Strideable. For compatibility, CountableRange is still available as a type alias.

There is one source compatibility issue with this: currently if you extend a CountableRange with an identical method/property as Range, you will get a compilation error of a duplicate. This shouldn't really happen – specifying CountableRange should be similar to Range where Bound: Strideable but generic type aliases need an enhancement to make this work.

GRDB is the one project that does this (Foundation also did similar things but has been changed to use RangeExpression instead, avoiding the issue). XFAILing this in the compatibility test suite until the type alias change has been made.

XFAIL ObjectMapper now that IUOs are removed from the type system and their use of overloading functions by different kinds of optionals in the case of inout parameters is something we can no longer represent.

[rudkx]

@swift-ci test

[rudkx]

@swift-ci test

[rudkx]

Hmm, perhaps this is actually a problem with something in my JSON.

[lplarson]

There are a few comma issues in your json:

"configuration": "Release" <-- missing trailing comma
 +        "xfail": {
 +          "compatibility": {
 +            "3.0": {
 +              "branch": {
 +                "master": "https://bugs.swift.org/browse/SR-6690" <-- missing trailing comma
 +                "swift-4.1-branch": "https://bugs.swift.org/browse/SR-6690"
 +              }
 +            }, <-- trailing comma that needs to be removed
 +	  }

[lplarson]

@swift-ci test

[lplarson]

@swift-ci test

[lplarson]

@rudkx Is this still needed? The compat jobs are passing.

[rudkx]

It isn't needed at the moment because I backed off on emitting an error with the compiler change I made, instead emitting a warning.

However, I have other changes in the works that will make it impossible to emit the warning, so we'll be back to needing to XFAIL until the project is updated to not have inout overloads that vary only in the kind of optional involved.

[rudkx]

We can close this out if you'd prefer, until that other change is ready to merge.

[rudkx]

I am ready to merge swiftlang/swift#14299, so we'll need to merge this in as well as XFAIL SwiftLint until its dependencies are updated to use a newer SourceKitten.

[rudkx]

/cc @lancep

[rudkx]

Please test with following PR:
swiftlang/swift#14299

@swift-ci Please test

[rudkx]

@swift-ci Please test

[rudkx]

The failure here appears due to swiftlang/swift#13342, for which we have #131 open to XFAIL GRDB.swift, but the latter failed.

[rudkx]

@swift-ci Please test with the following PR:
#131

@swift-ci Please test

[rudkx]

Okay, trying again, this time after merging #131 into my PR.

[rudkx]

@swift-ci Please test

[rudkx]

@swift-ci Please test

[rudkx]

@swift-ci Please test

[rudkx]

This is now blocked by a bunch of other failures: https://forums.swift.org/t/source-compatibility-bots-in-bad-shape/9526

[rudkx]

@swift-ci Please test and merge

[rudkx]

@swift-ci Please test