Don't use StaticString for XCTAssert*

[briancroom]

This addresses SR-308, bringing us in line with Apple's XCTest in using String instead of StaticString as the type of the file parameter on assertions.

Included is a minor refactor of how failure handling is structured in order to allow proper failure reporting before termination in the case of continueAfterFailure being false. I managed to reduce global state a little in the process.

[gribozavr]

Why not change the XCTest overlay instead? StaticString is inherently cheaper than String.

[modocache]

Why not change the XCTest overlay instead?

Oh, this would be awesome.

Slightly related: swiftlang/swift#785. It would be cool to expand the overlay's test suite so that it could catch any regressions in the StaticString change. swiftlang/swift#785 may be a useful reference when doing so.

[mike-ferris]

I agree with @gribozavr... I would rather see this redone as a pull request for the XCTest overlay in the swift project instead of this change.

[mike-ferris]

Actually I see you already did. I merged #888 in the swift project, so let's go ahead and close this one.

[briancroom]

@mike-ferris-apple how do you feel about the refactoring in briancroom@cc2ae5d ? Would you consider merging it if I open a separate PR?

I consider it an improvement due to the reduction of global vars and the removal of the public testFailure function.

[mike-ferris]

Ah, I didn’t notice that. Yeah, that seems like an improvement. It also gets rid of that weird public function that shouldn’t be public (and probably doesn’t need to be anymore, but I’ve never gotten around to verifying that…)

Go ahead and do a new PR for that if you want!

Mike

On Jan 7, 2016, at 4:35 PM, Brian Croom [email protected] wrote:

@mike-ferris-apple https://github.com/mike-ferris-apple how do you feel about the refactoring in briancroom/swift-corelibs-xctest@cc2ae5d briancroom@cc2ae5d ? Would you consider merging it if I open a separate PR?

I consider it an improvement due to the reduction of global vars and the removal of the public testFailure function.

—
Reply to this email directly or view it on GitHub #27 (comment).