Pete's Log: incomprehension
Entry #635, (Coding, Hacking, & CS stuff)(posted when I was 22 years old.)
oh man. So granted, it kinda feels like cheating when I use strace to figure out how tcsh does stuff. But on the other hand, how else am I supposed to figure out that it's perfectly normal for a process to receive a SIGTTOU when it sets the foreground process group of its controlling terminal to be its own process group? And that the simple solution to that problem is to simply ignore SIGTTOU while switching the foreground process group. I still don't know why this happens, but it makes things work. So now life is good. Almost. I'm approaching usable job control in my shell. Well, at least as far as infrastructure goes. Still need to implement some of them shell builtins that manipulate jobs. My main problem is that somehow I'm losing a SIGCONT that my shell parent is sending its children. And I don't know why! If I add debugging code around the kill system call, the signal gets sent. So I'm guessing we're hanging out in "race condition land" ... an unhappy place, to say the least. But it's all good, I'm watching Blues Brothers, drinking caffeine, and have a day full of football to look forward to tomorrow.