On 14/12/2018 09:47, Herbert Xu wrote:
On Fri, Dec 14, 2018 at 09:37:09AM +0000, Harald van Dijk wrote:
Still, that works as well, doesn't it? The control flow is basically the
same so the logic in my other message applies. returncmd() doesn't use
exceptions, so you get to dotrap() which already resets exitstatus if
necessary.
returncmd itself doesn't jump up but it may be part of a subshell
which would execute a jump to the top as part of the EV_EXIT
optimisation. That's where you'd need to restore the exit status.
Ah, okay, so for something like
trap 'false; (return); echo $?' EXIT
you want to remember that it's executed as part of a trap action and
print 0, not 1.
I actually want that to print 1, so for me it's not a problem. However,
my patch can be trivially modified to handle that: just have returncmd()
check savestatus the same way exitcmd() does. I think that also allows
merging SKIPFUNC and SKIPFUNCDEF, since you can then assume exitstatus
was always set appropriately, so it allows for some further reduction in
code.
