On 2016-04-16, Geoff Wozniak <ge...@wozniak.ca> wrote: > # zsh -c 'x=$(false); echo $?' > 0
I can reproduce this. > I've done this a few times and every time it's the same: on my > Thinkpad X200 the test produces '0' whereas any other machine I > install 5.9 on it produces '1'. Very odd. What I'm seeing here (with ktrace) should not differ between OpenBSD installs. The $(...) part is executed in a child process. After the parent reads an EOF from the pipe to the child, it does kill(child, 0). And that's where things diverge: * On OpenBSD, the kill() returns -1, errno ESRCH. The parent proceeds to print the exit code 0, and only afterwards(!) fetches the return status of the child with wait4(). * On FreeBSD or Linux, the kill() returns 0. The parent proceeds to fetch the return status of the child with wait4() and later prints the exit code 1. Apparently zsh expects kill(zombie, 0) to succeed, but this fails on OpenBSD. -- Christian "naddy" Weisgerber na...@mips.inka.de
