SourceKitService Consumes CPU and Grinds Xcode to a Halt

I resolved another issue that was causing SourceKitService use up to 13GB of memory...

I had String(format line with lots of arguments:

return String(format: "%d,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f,%.3f", samples.count,sum1.x,sum1.y,sum1.z,sum1.rx,sum1.ry,sum1.rz,sum2.x,sum2.y,sum2.z,sum2.rx,sum2.ry,sum2.rz,sum3.x,sum3.y,sum3.z,sum3.rx,sum3.ry,sum3.rz)

when replaced with this it worked fine (no memory build up and normal CPU consumption)

    var output: String = ""

    output += String(format: "%d,", samples.count)
    output += String(format: "%.3f,%.3f,%.3f,", sum1.x, sum1.y, sum1.z)
    output += String(format: "%.3f,%.3f,%.3f,", sum1.rx, sum1.ry, sum1.rz)
    output += String(format: "%.3f,%.3f,%.3f,", sum2.x, sum2.y, sum2.z)
    output += String(format: "%.3f,%.3f,%.3f,", sum2.rx, sum2.ry, sum2.rz)
    output += String(format: "%.3f,%.3f,%.3f,", sum3.x, sum3.y, sum3.z)
    output += String(format: "%.3f,%.3f,%.3f", sum3.rx, sum3.ry, sum3.rz)

    return output

I was seeing the problem because I was declaring an array with about 60 elements that looked like this:

let byteMap = [

["ECG" : (0,12)],
["PPG" : (12,3)],
["ECG" : (15,12)],
["PPG" : (27,3)],
["ECG" : (30,12)]

By explicitly annotating the type like this:

let byteMap : [String: (Int, Int)] = [

["ECG" : (0,12)],
["PPG" : (12,3)],
["ECG" : (15,12)],
["PPG" : (27,3)],
["ECG" : (30,12)],

I was able to make it stop. I think it must have something to do with Swift's type-inference and type-checking that makes it go into a loop when it encounters a longish array.

This was in Xcode 6.2. I also deleted the ModuleCache as described above and now everything is good.

Ran into this problem with Xcode 6.1.1 earlier this afternoon (not beta, official released version). I had been running some code on Playground and was suspecting that to be the cause. CPU was pegged to nearly 100%, and Xcode was unable to complete builds.

So here's what I did:

1. Opened "Activity Monitor", which showed SourceKitService as the main CPU hog.

2. Within "Activity Monitor", double-clicked on the SourceKitService and clicked on "Open Files and Ports" section, which showed it was working on files under the /Users/myname/Library/Developer/Xcode/DerivedData/ModuleCache/ directory for a specific folder.

3. Deleted the specified folder (from a command-line, using rm -rf). The cache is regenerated based on Can I safely delete contents of Xcode Derived data folder? .

4. Using Activity Monitor again, Force-Quit SourceKitServer. Saw the now-all-too-familiar sign within Xcode saying that SourceKitService had crashed (so that's why SourceKitService sounded familiar!).

5. Repeated step 3.

The Mac is peaceful, again. No data was lost and Xcode didn't even have to be restarted (which I had tried unsuccessfully). Bottom line is that ModuleCache seems to be getting SourceKitService in a loop and deleting the folder seems to fix it. Hope this works for you too.

Bootnote:

By the way, the cause for SourceKitService issue was that I had too long an array declaration in my Swift class. I had over 200 entries in an array. Reduced it to 30 and the error went away. So the issue may have arisen due to some kind of stack overflow in apple code (pun intended).

This problem happened like 10 times, 8 times it happened when I connected an actual device and didn't run through simulator.

I am not so sure if my solution is a good one, but for me I believe the problem was due to switching between simulator and an actual device. It may sound weird but it was as if it was creating interference between cache files.

What solved my problem:

• Clean Build Folder:( on Xcode) Alt + Shift + Command + K
• Reset Content and Settings: (on Simulator) Command + Shift + K.
• Waited a bit longer than normal and overload Xcode with constant clicks

So basically before you try to run on any new device, just delete any cache.

EDIT

I just had the problem without any device connection. I just quit Xcode and opened it again and the problem was gone. Not sure my guess is it could be some re-indexing issue after you fetch/pull merge new code.