[Pkg-octave-devel] Bug#871214: Bug#871214: Bug#871214: octave-image FTBFS on i386: test hangs

[Sébastien Villemot]

Le jeudi 10 août 2017 à 09:31 +0200, Rafael Laboissière a écrit :
> > * Adrian Bunk <bunk at debian.org> [2017-08-07 01:57]:
> 
> > Source: octave-image
> > Version: 2.6.1-1
> > Severity: serious
> > Tags: buster sid
> > 
> > Some recent change in unstable and buster makes octave-image FTBFS on i386:
> > 
> >  https://tests.reproducible-builds.org/debian/history/octave-image.html
> >  https://tests.reproducible-builds.org/debian/rb-pkg/unstable/i386/octave-image.html
> > 
> >  ...
> >  Checking CC files ...
> >  warning: function /build/1st/octave-image-2.6.1/inst/private/iscolormap.m shadows a core library function
> >  warning: called from
> >     /tmp/filePMKFm8 at line 1 column 1
> >  [__spatial_filtering__]
> >  PASSES 21 out of 21 tests
> >  [graycomatrix]
> >  PASSES 2 out of 2 tests
> >  [watershed]
> >  ***** test
> >  im = [
> >      2     3    30     2
> >      3    30     3    30
> >    255    31    30     4
> >      2   255    31    30
> >      1     2   255     5];
> > 
> >  labeled8 = [
> >      1     1     0     3
> >      1     1     0     3
> >      0     0     0     0
> >      2     2     0     4
> >      2     2     0     4];
> >  assert (watershed (im), labeled8);
> >  assert (watershed (im, 8), labeled8);
> >  !!!!! test failed
> >  ASSERT errors for:  assert (watershed (im),labeled8)
> > 
> >   Location  |  Observed  |  Expected  |  Reason
> >    (3,4)          3            0         Abs err 3 exceeds tol 0
> >    (4,4)          0            4         Abs err 4 exceeds tol 0
> >  [bwdist]
> >  Sat Sep  8 05:36:25 UTC 2018 - pbuilder was killed by timeout after 18h.
> 
> I would guess that the culprit is the following unit test in bwdist:
> 
>  ## The quasi-euclidean method is apparently sensitive to a machine precision
>  ## error that happens in x86 systems only. This test will cause an endless
>  ## loop in case of a regression.
>  %!test
>  %! bw = [  0   1   1   0   0   0   1   0
>  %!         0   0   0   0   0   0   0   0
>  %!         1   1   0   0   0   0   0   0
>  %!         0   0   0   0   0   0   1   0
>  %!         0   0   0   0   1   0   0   1
>  %!         0   0   0   0   0   0   0   0
>  %!         1   0   0   0   0   0   0   0
>  %!         0   0   1   0   0   1   1   0];
>  %! out = single ([
>  %! 1.00000   0.00000   0.00000   1.00000   2.00000   1.00000   0.00000   1.00000
>  %! 1.00000   1.00000   1.00000   sqrt(2)   sqrt(2)+1 sqrt(2)   1.00000   sqrt(2)
>  %! 0.00000   0.00000   1.00000   2.00000   2.00000   sqrt(2)   1.00000   sqrt(2)
>  %! 1.00000   1.00000   sqrt(2)   sqrt(2)   1.00000   1.00000   0.00000   1.00000
>  %! 2.00000   2.00000   2.00000   1.00000   0.00000   1.00000   1.00000   0.00000
>  %! 1.00000   sqrt(2)   2.00000   sqrt(2)   1.00000   sqrt(2)   sqrt(2)   1.00000
>  %! 0.00000   1.00000   1.00000   sqrt(2)   sqrt(2)   1.00000   1.00000   sqrt(2)
>  %! 1.00000   1.00000   0.00000   1.00000   1.00000   0.00000   0.00000   1.00000
>  %! ]);
>  %! assert (bwdist (bw, "quasi-euclidean"), out);
> 
> Should we run the test above only on non-x86 architectures?

I would rather patch the test to allow for some tolerance margin.

-- 
⢀⣴⠾⠻⢶⣦   Sébastien Villemot
⣾⠁⢠⠒⠀⣿⡁  Debian Developer
⢿⡄⠘⠷⠚⠋⠀  http://sebastien.villemot.name
⠈⠳⣄       http://www.debian.org