Re: 3.8.0-rc1 make check failures on Fedora rawhide

[Rik]

On 12/08/2013 10:00 AM, address@hidden wrote:
> Message: 1
> Date: Sat, 07 Dec 2013 11:26:29 -0700
> From: Orion Poplawski <address@hidden>
> To: Octave Maintainers List <address@hidden>
> Subject: 3.8.0-rc1 make check failures on Fedora rawhide
> Message-ID: <address@hidden>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
>  From Fedora rawhide scratch build:
>
> http://koji.fedoraproject.org/koji/taskinfo?taskID=6267556
>
>
>    libinterp/corefcn/graphics.cc-tst ...................... PASS   18/21 
>    FAIL 3
>    libinterp/dldfcn/chol.cc-tst ........................... PASS   28/29 
>    FAIL 1
>    scripts/gui/uimenu.m ................................... PASS    0/3 
>     FAIL 3
>    scripts/plot/util/allchild.m ........................... PASS    2/3 
>     FAIL 1
>    scripts/plot/util/findall.m ............................ PASS    0/1 
>     FAIL 1
>    scripts/plot/util/graphics_toolkit.m ................... PASS    0/2 
>     FAIL 2
>    bug-38236.tst .......................................... PASS    0/1 
>     FAIL 1
>
> Summary:
>    PASS     11446
>    FAIL        12
>    XFAIL        6
>    SKIPPED     43
>
> Most of these are coming from the lack of DISPLAY.  Shouldn't they be 
> skipped instead?
12/9/13

Orion,

The short answer is no.  When no DISPLAY variable is set Octave reverts to
using gnuplot which can make text-only plots.  But, the %!testif HAVE_FLTK
is checking a configure-time switch, whether Octave was built with FLTK. 
The test suite finds that you have FLTK and proceeds to try and switch the
graphic toolkit over to FLTK which then fails because no DISPLAY is set. 
We could be checking the return code of the call to graphics_toolkit or
wrapping these tests in unwind_protect blocks, but it doesn't feel like
it's worth the effort to track these down.  'make check' is used for sanity
checking the distribution and it seems easier to just arrange for the check
to be made on a non-headless system.

--Rik