Re: Crash on jobstates since workers/49783 (commit 6a8aa2a)

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Mon, Mar 28, 2022 at 7:49 AM Peter Stephenson
<p.w.stephenson@xxxxxxxxxxxx> wrote:
>
> Just thinking the pukka way to do this is actually zero the PN fields
> in the saved job structure when we enter the subshell (unless we decide
> to preserve them properly, which is more work but should be possible).

Looking at the way zargs was using subshells and $jobstates, I think
it's important that there be some way to distinguish jobs that the
subshell can't manage.

% % sleep 5 & ( sleep 2 & print -aC2 ${(kv)jobstates} )
[1] 68859
1  running:+
3  running:

How is it helpful that this returned anything?  Previously the
subshell could get information about the jobs it started itself, but
now it gets nothing.  I'm not even sure what job 3 is, there, as it
only shows up in the table at all a small fraction of the time when I
try this (usually it's just job 1).