Unicode characters kill the parsing of all subsequent // MARK tags

[brennon]

Using // MARK ... is super useful in breaking up jazzy's outputted documentation. With code such as this:

/**
    Returns the location of one of the gesture's touches in the local
    coordinate system of a given node.    
*/
func locationOfTouch(touchIndex index: Int, inNode node: SKNode?) -> CGPoint {
    return trackingTouches[index].locationInNode(node)
}

// MARK: Getting the Recognizer’s State and Node

I get delicious output. However, if I change that to this:

/**
    Returns the location of one of the gesture’s touches in the local
    coordinate system of a given node.    
*/
func locationOfTouch(touchIndex index: Int, inNode node: SKNode?) -> CGPoint {
    return trackingTouches[index].locationInNode(node)
}

// MARK: Getting the Recognizer’s State and Node

That // MARK and every one that follows is not parsed. The difference between the two chunks of source is hard to spot--it's the single quotation mark/inverted comma: ’ vs. '

To be clear, every // MARK before the non-ASCII/Unicode character is parsed. In addition, all of my other documentation is parsed. The problem is exactly and only this: any // MARK preceded in the original source by a non-ASCII character is not produced in the outputted documentation.

[brennon]

I just noticed this--the quotes in the // MARKs themselves can be non-ASCII (as they are in these two examples). The first example in my original report includes: "Getting the Recognizer’s State and Node" as the mark's text--that's a non-ASCII single quote in "Recognizer’s". I've double-checked that this doesn't cause any problems. The problem seems to be only with non-ASCII characters in the Headerdoc/Appledoc/reSTy (whatever we're calling them) documentation comments.

[jpsim]

Thanks for reporting this! I just pushed a fix to sourcekitten, which will be included in the next release.

[segiddins]

@jpsim did you add a small test case for this?