[PATCH util-modular v2] release.sh: unset GPGKEY if already existing

[Peter Hutterer]

On Thu, Apr 20, 2017 at 08:18:45AM +0300, Andres Gomez wrote:
> On Thu, 2017-04-20 at 10:17 +1000, Peter Hutterer wrote:
> > On Wed, Apr 19, 2017 at 03:59:33PM +0300, Andres Gomez wrote:
> > > GPGKEY may already exist in the environment. In that case, just
> > > unset it.
> > > 
> > > v2: unsetting is safer than redefining to quietly use a potentially
> > >     wrong key stored in the variable (Peter).
> > > 
> > > Signed-off-by: Andres Gomez <agomez at igalia.com>
> > > Cc: Emil Velikov <emil.velikov at collabora.com>
> > > Cc: Peter Hutterer <peter.hutterer at who-t.net>
> > > ---
> > >  release.sh | 5 +++++
> > >  1 file changed, 5 insertions(+)
> > > 
> > > diff --git a/release.sh b/release.sh
> > > index f976f94..b8a0aaf 100755
> > > --- a/release.sh
> > > +++ b/release.sh
> > > @@ -808,6 +808,11 @@ if [ "x$GPG" = "x" ] ; then
> > >      fi
> > >  fi
> > >  
> > > +# Unset GPGKEY if needed
> > > +if [ ! -z ${GPGKEY+x} ] ; then
> > > +    unset GPGKEY
> > > +fi
> > 
> > wouldn't that fail if GPGKEY is set to the empty string? Plus, I had to look
> > up what ${foo+x} actually does :) Should we just stick to the simple well
> > known:
> > 
> > if [ "x$GPGKEY" != "x" ] ; then
> >     unset GPGKEY
> > fi
> 
> ${foo+x} actually ensures it. That solution always unsets if the
> variable is set, empty string or not.

ok, that's good enough for me then, thanks, but...

> I suppose this other version would work too but, then, we will have a
> different outcome if GPGKEY is set with or without an empty string. In
> that case I would go for an even simpler initialization:
> 
> GPGKEY="" 
> 
> That way you ensure that GPGKEY is always set to an empty string.
> 
> Let me know what you prefer.

.. let's do the unconditional unset please, it's the most expressive one
and skips the extra checks. sorry about the churn.

Cheers,
   Peter