Re: Failing Java tests due to no JRE

[John W. Eaton]

On 09/15/2016 01:29 PM, Rik wrote:
> On 09/15/2016 08:12 AM, Mike Miller wrote:
> > On Wed, Sep 14, 2016 at 14:59:59 -0700, Rik wrote:
> > > I'm testing the 4.2.0-rc1 candidate on a Windows XP virtual machine which
> > > does not have Java installed.  When using __run_test_suite__ all of the
> > > Java tests fail (Duh!).  The issue is that all of the tests begin with
> > >
> > > %!testif HAVE_JAVA
> > >
> > > Certainly, when Octave was cross-compiled with MXE Java was available, but
> > > now at runtime there is no Java environment.  Should all of these tests
> > > begin with
> > >
> > > %!testif HAVE_JAVA
> > > %! if (! usejava ("jvm"))
> > > %!   return;
> > >
> > > so that if no JVM is available at run time they are not executed?
> > Yes I think that's a good idea. It's also possible to run Octave on any
> > GNU or Unix system with no JRE installed and the same test failure would
> > happen there.
>
> Done.  See cset http://hg.savannah.gnu.org/hgweb/octave/rev/a3f74fed899b.

That seems OK as a way to avoid failures but doesn't it also cause the 
test to succeed since the test will return early with no error? 
Shouldn't it report "skipped" instead?  If so, I think we'll need to 
enhance the way the condition in a testif block works.  But that is 
definitely not urgent, so can wait until after the release.

Thanks,

jwe