Interrupting carthage during build doesn't completely interrupt about carthage HOT 7 CLOSED

This is specifically with xcodebuild, right? I've long suspected that xcodebuild does some interesting things to avoid quitting immediately—perhaps we should SIGKILL child processes on exit.

from carthage.

Yep. I tried the "Moya" experiment again and killed it when it started spitting DVT exceptions "please file a radar" shit everywhere. It did it about three times total, so perhaps number of logical cores.

from carthage.

Thanks. I'll take a look into it at some point.

In the meantime, the workaround I usually use is hitting Cmd-Z (interrupt), then running kill %1 to forcibly kill the interrupted process.

from carthage.

Cheers – more you know!

from carthage.

From a lot of Googling and consulting with @keithduncan, it appears that NSTask is a special 💩❄️ in this scenario, and spawns processes outside of the parent's process group. The same problem supposedly wouldn't arise with fork() or posix_spawn() directly.

from carthage.

This is pretty terrible, but it seems like our only options are:

1. Create a dispatch_source_t for each signal we wish to handle, then forward those to spawned child processes.
2. Create an NSTask replacement based on fork() or posix_spawn().

Consequently, I'm gonna avoid shaving this yak for now. 👢

from carthage.

Moved to Carthage/ReactiveTask#3.

from carthage.