Re: Not quite out of the pipestatus woods yet ...

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Oct 25,  4:25pm, Bart Schaefer wrote:
}
} setopt MONITOR
} 
} before the "stress test" in A05execution.ztst, it ends up failing most of
} the time.  So there is indeed still a race of some sort.

Some experimentation with "set -x" and TRAPCHLD indicates that it gets
pipestatus wrong if the left-hand-sides of the pipe exit while the
loop body is still running.

Some debugging statements added to storepipestats() indicate that the
problem is that job != thisjob at the time the loop finally does exit.

I think it has something to do with the oldjobtab / jobtab switching, but
have not yet confirmed.