On Fri, 6 Nov 2015 16:21:03 +0100 Andreas Henriksson <andr...@fatal.se> wrote: > Hi again. > > On Fri, Nov 06, 2015 at 03:01:08PM +0100, Peter Palfrader wrote: > [...] > > But it should work. > > > > An "exit 0" at the end, or a "if [ -d ... ]; then rmdir .. ; fi" would > > also work instead and might be preferable. > > Please stop suggesting exit 0. It does not work. Hopefully the attached > minimal testcase will convince you of this. The final "exit 0" will > simply never be reached. run: ./foo.sh && echo $? > (Try also replace 'false' with 'false && true' to more exactly simulate > your particular bug case.) > > (... and even if it was reached, that would certainly throw away > any exit code - not just the rmdir one.) > > rmdir failing is not the end of the world anyway, it's just a > "lets be nice and clean up if we can" kind of thing. > > In your very obscure case that noone else triggered so far there > is no directory to clean up anyway, so don't worry! :P
For the record, I've triggered this today when upgrading an aarch64 chroot. Maybe it's not that obscure ;) Cheers, Javi