Dear Ludo’,

many thanks for the feedback! I have fixed the minor issues you
mentioned.

On Mon, Jun 19, 2017 at 02:14:29PM +0200, Ludovic Courtès wrote:
> > +    (outputs '("out" "python"))
> > +    (inputs
> > +     `(("expat" ,expat)
> > +       ("flex" ,flex)
> > +       ("libevent" ,libevent)
> > +       ("protobuf" ,protobuf)
> > +       ("python" ,python-3)
> > +       ("python-wrapper" ,python-wrapper)
> > +       ("openssl" ,openssl)
> > +       ("swig" ,swig)))
> 
> I think SWIG should go to ‘native-inputs’ because it’s only used at
> build time.
Right, and I also moved ‘flex’ there. 

> > +             (let* ((source (assoc-ref %build-inputs "source"))
> > +                    (gcc (assoc-ref %build-inputs "gcc")))
> > +               (call-with-output-file "/tmp/nss_preload.c"
> > +                 (lambda (port)
> > +                   (display "#include <stdlib.h>
> 
> As discussed on help-guix, I would prefer using ‘substitute*’ to modify
> all the unit tests.  That would reduce complexity and be potentially
> more robust.  Does that sound feasible without much hassle?
Substituting is not not an option since after changing a record in the
test data one would have to create (cryptographically) new matching RRSIG 
records otherwise the test would still fail.

> That’s it, thank you for this first package!
Thanks again for your help! Attached is the modified patch.

Best,
Gregor