Re: [PATCH v3] iotests: Drop readlink -f

[Peter Maydell]

On Mon, 14 Sep 2020 at 17:03, Eric Blake <eblake@redhat.com> wrote:
> If CDPATH is set, this can produce wrong results.  Do we care?  Probably
> not, since (as you point out), $source_iotests has the same bug, and no
> one has complained yet.

This has come up previously -- configure and probably most
of our other shell scripts will also break in the presence
of CDPATH. I stand by my position that CDPATH is a bad
misfeature that the shell should not be applying to
scripts. If anybody ever reports a bug we could I suppose
unconditionally unset CDPATH at the top of all our scripts.

thanks
-- PMM