Closed jamztang closed 6 years ago
@jamztang Thank you for providing the stack trace. I'm looking into this. Are you able to reproduce this consistently? If so, can you provide steps to reproduce it?
Actually we're not 100% sure how to reproduce this, but this happens to be 2 of our most happening crashes
One info is that our app is a OSX menu bar application, it could go active and resignActive pretty quickly. If I add logs to AppDelegate and LDEventSource I see something like this.
AppDelegate.willBecomeActive
LDEventSource._open
AppDelegate.willResignActive
LDEventSource.close
LDEventSource.receiveResponse // Sometimes it doesn't but sometimes it shows.
We're trying to disable streaming
for now and see how it goes.
Knowing that this is a macOS app is helpful. Also knowing that the app goes to the background quickly is helpful. The LDEventSource delays making the connection to the clientstream for 1 second. If the app moves to the background before that time, then the SDK should dealloc the LDEventSource. I will experiment with our sample macOS app to see if I can duplicate the crash there.
After studying these for some time, I have a couple of more questions. Do you use JavaScript in your app? Or maybe a JS based library? If so, do you setup any timers, or set a handler on a timer provided to you by a third party?
I appreciate any more information you can give us.
Hi @jamztang , We believe this issue is fixed in iOS SDK 2.12.
Hi guys, I was tracking down a crash, and seems like it's related to DarklyEventSource.
I went further to look at our Podfile.lock, we're referencing to LaunchDarkly 2.10.1 and DarklyEventSource 3.2.0
The closest thing I can find is this: https://github.com/launchdarkly/ios-eventsource/blob/3.2.0/LDEventSource/LDEventSource.m#L293-L295
Are there any known reports that CFRunLoopRun will crash is certain situation?