Re: Failures from __ode15__ in 'make check'

[John W. Eaton]

On 12/17/2016 04:12 PM, Rik wrote:
> On 12/17/2016 12:58 AM, Carlo De Falco wrote:
> > On 16 Dec 2016, at 22:06, Rik <address@hidden> wrote:
> >
> > > 12/16/16
> > >
> > > Carlo,
> > >
> > > I'm getting 50+ failures from __ode15__.cc when running 'make check'.  They
> > > are all of this type
> > >
> > > ***** test  # Mass option as matrix
> > > opt = odeset ("Mass", eye (2,2), "MStateDependence", "none");
> > > sol = ode15s (@fpol, [0 2], [2 0], opt);
> > > assert ([sol.x(end), sol.y(end,:)], [2, fref], 3e-3);
> > > !!!!! test failed
> > > __ode15__: support for sundials_ida, sundials_nvecserial was unavailable or
> > > disabled when Octave was built
> > >
> > > It appears that a common declaration, like HAVE_SUNDIALS, needs to be
> > > declared by configure so that  these tests are only run when there is
> > > support for them.  The correstponding tests can then begin with
> > >
> > > %!testif HAVE_SUNDIALS
> > >
> > > --Rik
> >
> > Rik,
> >
> > Thanks for the feedback.
> >
> > There is already such a flag defined in the config files,
> > is there anything else that needs to be done or can I just
> > add %!testif HAVE_SUNDIALS ?
>
> No, that's all there is to it.

I checked in a changeset that fixes the %!test blocks.

I'm not sure what to do about the %!error blocks.  Currently they pass 
because there is no text to match, but I suspect that at least some of 
them will start to fail if you add expected error messages because the 
actual error message may be different when Sundials is present vs when 
it is missing.  We don't have an %!errorif block, and I'm not really 
excited by the idea of adding it.

jwe