No way to stop a BridgeFinder scan

[ahamel]

I can't seem to find a way to actually stop the scanner during the BridgeFinder flow. For example if you wanted to stop the scan with some timeout etc... I don't see any public API on the BridgeFinder class.

[NilsLattek]

Right now there are two ways implemented to scan for bridges: SSDPScanner which searches inside the local network and NUPNPScanner which makes a REST call to the Hue-cloud to receive the devices.
The first solution uses a scan timeout of 5 seconds:

SwiftyHue/Sources/BridgeServices/BridgeFinder/Scanner/SSDPScanner.swift

Line 37 in b004b5c

let receiveTimeout: TimeInterval = 5

and the second method is a standard HTTP request with the standard timeout of 60 seconds

SwiftyHue/Sources/BridgeServices/BridgeFinder/Scanner/NUPNPScanner.swift

Line 20 in b004b5c

let request = createRequest()

This is just for finding the bridge IP after that it also needs to call each IP to validate that it is a valid bridge which is done with a REST call again:

SwiftyHue/Sources/BridgeServices/BridgeFinder/Validator/BridgeValidator.swift

Line 22 in b004b5c

let request = NSMutableURLRequest(url: url)

which also uses the default timeout. This call is done for each IP it can find during the first step.
I guess it would make sense to reduce the HTTP-timeouts a litte bit to 10-15seconds or something like that.
So right now it is not scanning forever.

Adding a stop() method is also an option and it is already implemented in the first scanner:

SwiftyHue/Sources/BridgeServices/BridgeFinder/Scanner/SSDPScanner.swift

Line 45 in b004b5c

@objc func stop() {

and supported in the scanner interface.
The remaining work would be to also cancel the request in the NUPNPScanner scanner and for the validator requests.

[ahamel]

Thanks for the detailed update. Yeah I was looking deeper into the docs for Hue. It actually looks like they now prefer mDNS for discovery. I actually implemented a mDNS scanner in this architecture as an additional scanner. I did see that there are several scanners at play, prioritized by preferred scanning methods first. Which would make stop() or cancel() a little tricky. I was just thinking that it would be nice to cleanup the scanners if the user decided to stop waiting for scanning. etc...

[NilsLattek]

As a workaround: I guess you could just remove the delegate and dealloc the object. It would still run in the background until the internal timeout is finished and then it is removed, but it won't call the delegate methods. Not the nicest way though, but should work.
Feel free to create a PR which implements the missing stop functions for the requests or new scanners. We would be happy to have some help.